conference
stringclasses 6
values | title
stringlengths 2
176
| abstract
stringlengths 2
5k
| decision
stringclasses 11
values |
|---|---|---|---|
ICLR.cc/2020/Conference | Composing Task-Agnostic Policies with Deep Reinforcement Learning | The composition of elementary behaviors to solve challenging transfer learning problems is one of the key elements in building intelligent machines. To date, there has been plenty of work on learning task-specific policies or skills but almost no focus on composing necessary, task-agnostic skills to find a solution to new problems. In this paper, we propose a novel deep reinforcement learning-based skill transfer and composition method that takes the agent's primitive policies to solve unseen tasks. We evaluate our method in difficult cases where training policy through standard reinforcement learning (RL) or even hierarchical RL is either not feasible or exhibits high sample complexity. We show that our method not only transfers skills to new problem settings but also solves the challenging environments requiring both task planning and motion control with high data efficiency. | Accept (Poster) |
ICLR.cc/2020/Conference | ILS-SUMM: Iterated Local Search for Unsupervised Video Summarization | In recent years, there has been an increasing interest in building video summarization tools, where the goal is to automatically create a short summary of an input video that properly represents the original content. We consider shot-based video summarization where the summary consists of a subset of the video shots which can be of various lengths. A straightforward approach to maximize the representativeness of a subset of shots is by minimizing the total distance between shots and their nearest selected shots. We formulate the task of video summarization as an optimization problem with a knapsack-like constraint on the total summary duration. Previous studies have proposed greedy algorithms to solve this problem approximately, but no experiments were presented to measure the ability of these methods to obtain solutions with low total distance. Indeed, our experiments on video summarization datasets show that the success of current methods in obtaining results with low total distance still has much room for improvement. In this paper, we develop ILS-SUMM, a novel video summarization algorithm to solve the subset selection problem under the knapsack constraint. Our algorithm is based on the well-known metaheuristic optimization framework -- Iterated Local Search (ILS), known for its ability to avoid weak local minima and obtain a good near-global minimum. Extensive experiments show that our method finds solutions with significantly better total distance than previous methods. Moreover, to indicate the high scalability of ILS-SUMM, we introduce a new dataset consisting of videos of various lengths. | Withdrawn |
ICLR.cc/2019/Conference | DEEP-TRIM: REVISITING L1 REGULARIZATION FOR CONNECTION PRUNING OF DEEP NETWORK | State-of-the-art deep neural networks (DNNs) typically have tens of millions of parameters, which might not fit into the upper levels of the memory hierarchy, thus increasing the inference time and energy consumption significantly, and prohibiting their use on edge devices such as mobile phones. The compression of DNN models has therefore become an active area of research recently, with \emph{connection pruning} emerging as one of the most successful strategies. A very natural approach is to prune connections of DNNs via $\ell_1$ regularization, but recent empirical investigations have suggested that this does not work as well in the context of DNN compression. In this work, we revisit this simple strategy and analyze it rigorously, to show that: (a) any \emph{stationary point} of an $\ell_1$-regularized layerwise-pruning objective has its number of non-zero elements bounded by the number of penalized prediction logits, regardless of the strength of the regularization; (b) successful pruning highly relies on an accurate optimization solver, and there is a trade-off between compression speed and distortion of prediction accuracy, controlled by the strength of regularization. Our theoretical results thus suggest that $\ell_1$ pruning could be successful provided we use an accurate optimization solver. We corroborate this in our experiments, where we show that simple $\ell_1$ regularization with an Adamax-L1(cumulative) solver gives pruning ratio competitive to the state-of-the-art. | Reject |
ICLR.cc/2022/Conference | AF$_2$: Adaptive Focus Framework for Aerial Imagery Segmentation | As a specific semantic segmentation task, aerial imagery segmentation has been widely employed in high spatial resolution (HSR) remote sensing images understanding. Besides common issues (e.g. large scale variation) faced by general semantic segmentation tasks, aerial imagery segmentation has some unique challenges, the most critical one among which lies in foreground-background imbalance. There have been some recent efforts that attempt to address this issue by proposing sophisticated neural network architectures, since they can be used to extract informative multi-scale feature representations and increase the discrimination of object boundaries. Nevertheless, many of them merely utilize those multi-scale representations in ad-hoc measures but disregard the fact that the semantic meaning of objects with various sizes could be better identified via receptive fields of diverse ranges. In this paper, we propose Adaptive Focus Framework (AF$_2$), which adopts a hierarchical segmentation procedure and focuses on adaptively utilizing multi-scale representations generated by widely adopted neural network architectures. Particularly, a learnable module, called Adaptive Confidence Mechanism (ACM), is proposed to determine which scale of representation should be used for the segmentation of different objects. Comprehensive experiments show that AF$_2$ has significantly improved the accuracy on three widely used aerial benchmarks, as fast as the mainstream method.
| Reject |
ICLR.cc/2022/Conference | INFERNO: Inferring Object-Centric 3D Scene Representations without Supervision | We propose INFERNO, a method to infer object-centric representations of visual scenes without relying on annotations. Our method learns to decompose a scene into multiple objects, each object having a structured representation that disentangles its shape, appearance and 3D pose. To impose this structure we rely on recent advances in neural 3D rendering. Each object representation defines a localized neural radiance field that is used to generate 2D views of the scene through a differentiable rendering process. Our model is subsequently trained by minimizing a reconstruction loss between inputs and corresponding rendered scenes. We empirically show that INFERNO discovers objects in a scene without supervision. We also validate the interpretability of the learned representations by manipulating inferred scenes and showing the corresponding effect in the rendered output. Finally, we demonstrate the usefulness of our 3D object representations in a visual reasoning task using the CATER dataset. | Reject |
ICLR.cc/2018/Conference | Federated Learning: Strategies for Improving Communication Efficiency | Federated Learning is a machine learning setting where the goal is to train a high-quality centralized model while training data remains distributed over a large number of clients each with unreliable and relatively slow network connections. We consider learning algorithms for this setting where on each round, each client independently computes an update to the current model based on its local data, and communicates this update to a central server, where the client-side updates are aggregated to compute a new global model. The typical clients in this setting are mobile phones, and communication efficiency is of the utmost importance.
In this paper, we propose two ways to reduce the uplink communication costs: structured updates, where we directly learn an update from a restricted space parametrized using a smaller number of variables, e.g. either low-rank or a random mask; and sketched updates, where we learn a full model update and then compress it using a combination of quantization, random rotations, and subsampling before sending it to the server. Experiments on both convolutional and recurrent networks show that the proposed methods can reduce the communication cost by two orders of magnitude. | Reject |
ICLR.cc/2023/Conference | Efficient and Stealthy Backdoor Attack Triggers are Close at Hand | A backdoor attack aims to inject a backdoor into a deep model so that the model performs normally on benign samples while maliciously predicting the input as the attacker-defined target class when the backdoor is activated by a predefined trigger pattern. Most existing backdoor attacks use a pattern that rarely occurs in benign data as the trigger pattern. In this way, the impact of the attack on the label prediction of benign data can be mitigated. However, this practice also results in the attack being defended against with little performance degradation on benign data by preventing the trigger pattern from being activated. In this work, we present a new attack strategy to solve this dilemma. Unlike the conventional strategy, our strategy extracts the trigger pattern from benign training data, which frequently occurs in samples of the target class but rarely occurs in samples of the other classes. Compared with the prevailing strategy, our proposed strategy has two advantages. First, it can improve the efficiency of the attack because learning on benign samples of the target class can facilitate the fitting of the trigger pattern. Second, it increases the difficulty or cost of identifying the trigger pattern and preventing its activation, since many benign samples of the target class contain the trigger pattern. We empirically evaluate our strategy on four benchmark datasets. The experimental studies show that attacks performed with our strategy can achieve much better performance when poisoning only 0.1\% or more of the training data, and can achieve better performance against several benchmark defense algorithms. | Reject |
ICLR.cc/2021/Conference | Once Quantized for All: Progressively Searching for Quantized Compact Models | Automatic search of Quantized Neural Networks (QNN) has attracted a lot of attention. However, the existing quantization-aware Neural Architecture Search (NAS) approaches inherit a two-stage search-retrain schema, which is not only time-consuming but also adversely affected by the unreliable ranking of architectures during the search. To avoid the undesirable effect of the search-retrain schema, we present Once Quantized for All (OQA), a novel framework that searches for quantized compact models and deploys their quantized weights at the same time without additional post-process. While supporting a huge architecture search space, our OQA can produce a series of quantized compact models under ultra-low bit-widths(e.g. 4/3/2 bit). A progressive bit inheritance procedure is introduced to support ultra-low bit-width. Our searched model family, OQANets, achieves a new state-of-the-art (SOTA) on quantized compact models compared with various quantization methods and bit-widths. In particular, OQA2bit-L achieves 64.0\% ImageNet Top-1 accuracy, outperforming its 2 bit counterpart EfficientNet-B0@QKD by a large margin of 14\% using 30\% less computation cost. | Reject |
ICLR.cc/2022/Conference | Adaptive Graph Capsule Convolutional Networks | In recent years, many studies utilize Convolutional Neural Networks (CNNs) to deal with non-grid graph data, known as Graph Convolutional Networks (GCNs). However, there exist two main restrictions of the prevalent GCNs. First, GCNs have a latent information loss problem since they use scalar-valued neurons rather than vector-valued ones to iterate through graph convolutions. Second, GCNs are presented statically with fixed architectures during training, which would limit their representation power. To tackle these two issues, based on a GNN model (CapsGNN) which encodes node embeddings as vectors, we propose Adaptive Graph Capsule Convolutional Networks (AdaGCCN) to adaptively adjust the model architecture at runtime. Specifically, we leverage Reinforcement Learning (RL) to design an assistant module for continuously selecting the optimal modification to the model structure through the whole training process. Moreover, we determine the architecture search space through analyzing the impacts of model's depth and width. To mitigate the computation overhead brought by the assistant module, we then deploy multiple workers to compute in parallel on GPU. Evaluations show that AdaGCCN achieves SOTA accuracy results and outperforms CapsGNN almost on all datasets in both bioinformatics and social fields. We also conduct experiments to indicate the efficiency of the paralleling strategy. | Withdrawn |
ICLR.cc/2023/Conference | Backdoors Stuck At The Frontdoor: Multi-Agent Backdoor Attacks That Backfire | Malicious agents in collaborative learning and outsourced data collection threaten the training of clean models. Backdoor attacks, where an attacker poisons a model during training to successfully achieve targeted misclassification, are a major concern to train-time robustness. In this paper, we investigate a multi-agent backdoor attack scenario, where multiple attackers attempt to backdoor a victim model simultaneously. A consistent backfiring phenomenon is observed across a wide range of games, where agents suffer from a low collective attack success rate. We examine different modes of backdoor attack configurations, non-cooperation / cooperation, joint distribution shifts, and game setups to return an equilibrium attack success rate at the lower bound. The results motivate the re-evaluation of backdoor defense research for practical environments. | Reject |
ICLR.cc/2023/Conference | Accelerated Single-Call Methods for Constrained Min-Max Optimization | We study first-order methods for constrained min-max optimization. Existing methods either require two gradient calls or two projections in each iteration, which may be costly in some applications. In this paper, we first show that a variant of the \emph{Optimistic Gradient (OG)} method, a \emph{single-call single-projection} algorithm, has $O(\frac{1}{\sqrt{T}})$ best-iterate convergence rate for inclusion problems with operators that satisfy the weak Minty variation inequality (MVI). Our second result is the first single-call single-projection algorithm -- the \emph{Accelerated Reflected Gradient (ARG)} method that achieves the \emph{optimal $O(\frac{1}{T})$} last-iterate convergence rate for inclusion problems that satisfy negative comonotonicity. Both the weak MVI and negative comonotonicity are well-studied assumptions and capture a rich set of non-convex non-concave min-max optimization problems. Finally, we show that the \emph{Reflected Gradient (RG)} method, another \emph{single-call single-projection} algorithm, has $O(\frac{1}{\sqrt{T}})$ last-iterate convergence rate for constrained convex-concave min-max optimization, answering an open problem of [Hsieh et al., 2019]. Our convergence rates hold for standard measures such as the tangent residual and the natural residual. | Accept: poster |
ICLR.cc/2021/Conference | Cross-Attentional Audio-Visual Fusion for Weakly-Supervised Action Localization | Temporally localizing actions in videos is one of the key components for video understanding. Learning from weakly-labeled data is seen as a potential solution towards avoiding expensive frame-level annotations. Different from other works which only depend on visual-modality, we propose to learn richer audiovisual representation for weakly-supervised action localization. First, we propose a multi-stage cross-attention mechanism to collaboratively fuse audio and visual features, which preserves the intra-modal characteristics. Second, to model both foreground and background frames, we construct an open-max classifier that treats the background class as an open-set. Third, for precise action localization, we design consistency losses to enforce temporal continuity for the action class prediction, and also help with foreground-prediction reliability. Extensive experiments on two publicly available video-datasets (AVE and ActivityNet1.2) show that the proposed method effectively fuses audio and visual modalities, and achieves the state-of-the-art results for weakly-supervised action localization. | Accept (Poster) |
ICLR.cc/2023/Conference | Rethinking the Expressive Power of GNNs via Graph Biconnectivity | Designing expressive Graph Neural Networks (GNNs) is a central topic in learning graph-structured data. While numerous approaches have been proposed to improve GNNs with respect to the Weisfeiler-Lehman (WL) test, for most of them, there is still a lack of deep understanding of what additional power they can systematically and provably gain. In this paper, we take a fundamentally different perspective to study the expressive power of GNNs beyond the WL test. Specifically, we introduce a novel class of expressivity metrics via graph biconnectivity and highlight their importance in both theory and practice. As biconnectivity can be easily calculated using simple algorithms that have linear computational costs, it is natural to expect that popular GNNs can learn it easily as well. However, after a thorough review of prior GNN architectures, we surprisingly find that most of them are not expressive for any of these metrics. The only exception is the ESAN framework (Bevilacqua et al., 2022), for which we give a theoretical justification of its power. We proceed to introduce a principled and more efficient approach, called the Generalized Distance Weisfeiler-Lehman (GD-WL), which is provably expressive for all biconnectivity metrics. Practically, we show GD-WL can be implemented by a Transformer-like architecture that preserves expressiveness and enjoys full parallelizability. A set of experiments on both synthetic and real datasets demonstrates that our approach can consistently outperform prior GNN architectures. | Accept: notable-top-5% |
ICLR.cc/2019/Conference | K for the Price of 1: Parameter-efficient Multi-task and Transfer Learning | We introduce a novel method that enables parameter-efficient transfer and multi-task learning with deep neural networks. The basic approach is to learn a model patch - a small set of parameters - that will specialize to each task, instead of fine-tuning the last layer or the entire network. For instance, we show that learning a set of scales and biases is sufficient to convert a pretrained network to perform well on qualitatively different problems (e.g. converting a Single Shot MultiBox Detection (SSD) model into a 1000-class image classification model while reusing 98% of parameters of the SSD feature extractor). Similarly, we show that re-learning existing low-parameter layers (such as depth-wise convolutions) while keeping the rest of the network frozen also improves transfer-learning accuracy significantly. Our approach allows both simultaneous (multi-task) as well as sequential transfer learning. In several multi-task learning problems, despite using much fewer parameters than traditional logits-only fine-tuning, we match single-task performance.
| Accept (Poster) |
ICLR.cc/2021/Conference | Response Modeling of Hyper-Parameters for Deep Convolutional Neural Networks | Hyper-parameter optimization (HPO) is critical in training high performing Deep Neural Networks (DNN). Current methodologies fail to define an analytical response surface and remain a training bottleneck due to their use of additional internal hyper-parameters and lengthy evaluation cycles. We demonstrate that the low-rank factorization of the convolution weights of intermediate layers of a CNN can define an analytical response surface. We quantify how this surface acts as an auxiliary to optimizing training metrics. We introduce a dynamic tracking algorithm -- autoHyper -- that performs HPO on the order of hours for various datasets including ImageNet and requires no manual tuning. Our method -- using a single RTX2080Ti -- is able to select a learning rate within 59 hours for AdaM on ResNet34 applied to ImageNet and improves in testing accuracy by 4.93% over the default learning rate. In contrast to previous methods, we empirically prove that our algorithm and response surface generalize well across model, optimizer, and dataset selection removing the need for extensive domain knowledge to achieve high levels of performance. | Reject |
ICLR.cc/2018/Conference | Achieving morphological agreement with Concorde | Neural conversational models are widely used in applications like personal assistants and chat bots. These models seem to give better performance when operating on word level. However, for fusion languages like French, Russian and Polish vocabulary size sometimes become infeasible since most of the words have lots of word forms. We propose a neural network architecture for transforming normalized text into a grammatically correct one. Our model efficiently employs correspondence between normalized and target words and significantly outperforms character-level models while being 2x faster in training and 20\% faster at evaluation. We also propose a new pipeline for building conversational models: first generate a normalized answer and then transform it into a grammatically correct one using our network. The proposed pipeline gives better performance than character-level conversational models according to assessor testing. | Reject |
ICLR.cc/2021/Conference | An Empirical Exploration of Open-Set Recognition via Lightweight Statistical Pipelines | Machine-learned safety-critical systems need to be self-aware and reliably know their unknowns in the open-world. This is often explored through the lens of anomaly/outlier detection or out-of-distribution modeling. One popular formulation is that of open-set classification, where an image classifier trained for 1-of-$K$ classes should also recognize images belonging to a $(K+1)^{th}$ "other" class, not present in the training set. Recent work has shown that, somewhat surprisingly, most if not all existing open-world methods do not work well on high-dimensional open-world images (Shafaei et al. 2019). In this paper, we carry out an empirical exploration of open-set classification, and find that combining classic statistical methods with carefully computed features can dramatically outperform prior work. We extract features from off-the-shelf (OTS) state-of-the-art networks for the underlying $K$-way closed-world task. We leverage insights from the retrieval community for computing feature descriptors that are low-dimensional (via pooling and PCA) and normalized (via L2-normalization), enabling the modeling of training data densities via classic statistical tools such as kmeans and Gaussian Mixture Models (GMMs). | Reject |
ICLR.cc/2023/Conference | Beyond re-balancing: distributionally robust augmentation against class-conditional distribution shift in long-tailed recognition | As a fundamental and practical problem, long-tailed recognition has drawn burning attention. In this paper, we investigate an essential but rarely noticed issue in long-tailed recognition, Class-Conditional Distribution (CCD) shift due to scarce instances, which exhibits a significant discrepancy between the empirical CCDs for training and test data, especially for tail classes. We observe empirical evidence that the shift is a key factor that limits the performance of existing long-tailed learning methods, and provide novel understanding of these methods in the
course of our analysis. Motivated by this, we propose an adaptive data augmentation method, Distributionally Robust Augmentation (DRA), to learn models more robust to CCD shift. The convergence and generalization of DRA are theoretically guaranteed. Experimental results verify that DRA outperforms related data augmentation methods without extra training cost and significantly improves the performance of some existing long-tailed recognition methods. | Reject |
ICLR.cc/2022/Conference | Case-based reasoning for better generalization in textual reinforcement learning | Text-based games (TBG) have emerged as promising environments for driving research in grounded language understanding and studying problems like generalization and sample efficiency. Several deep reinforcement learning (RL) methods with varying architectures and learning schemes have been proposed for TBGs. However, these methods fail to generalize efficiently, especially under distributional shifts. In a departure from deep RL approaches, in this paper, we propose a general method inspired by case-based reasoning to train agents and generalize out of the training distribution. The case-based reasoner collects instances of positive experiences from the agent's interaction with the world and later reuses the collected experiences to act efficiently. The method can be used in conjunction with any existing on-policy neural agent introduced in the literature for TBGs. Our experiments show that the proposed approach consistently improves existing methods, obtains good out-of-distribution generalization and achieves new state-of-the-art results on widely used environments. | Accept (Poster) |
ICLR.cc/2022/Conference | Adaptive Inertia: Disentangling the Effects of Adaptive Learning Rate and Momentum | Adaptive Momentum Estimation (Adam), which combines Adaptive Learning Rate and Momentum, would be the most popular stochastic optimizer for accelerating the training of deep neural networks. However, empirically Adam often generalizes worse than Stochastic Gradient Descent (SGD). We unveil the mystery of this behavior in the diffusion theoretical framework. Specifically, we disentangle the effects of Adaptive Learning Rate and Momentum of the Adam dynamics on saddle-point escaping and flat minima selection. We prove that Adaptive Learning Rate can escape saddle points efficiently, but cannot select flat minima as SGD does. In contrast, Momentum provides a drift effect to help the training process pass through saddle points, and almost does not affect flat minima selection. This partly explains why SGD (with Momentum) generalizes better, while Adam generalizes worse but converges faster. Furthermore, motivated by the analysis, we design a novel adaptive optimization framework named Adaptive Inertia, which uses parameter-wise adaptive inertia to accelerate the training and provably favors flat minima as well as SGD. Our extensive experiments demonstrate that the proposed adaptive inertia method can generalize significantly better than SGD and conventional adaptive gradient methods. | Reject |
ICLR.cc/2022/Conference | Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap | Recently, contrastive learning has risen to be a promising approach for large-scale self-supervised learning. However, theoretical understanding of how it works is still unclear. In this paper, we propose a new guarantee on the downstream performance without resorting to the conditional independence assumption that is widely adopted in previous work but hardly holds in practice. Our new theory hinges on the insight that the support of different intra-class samples will become more overlapped under aggressive data augmentations, thus simply aligning the positive samples (augmented views of the same sample) could make contrastive learning cluster intra-class samples together. Based on this augmentation overlap perspective, theoretically, we obtain asymptotically closed bounds for downstream performance under weaker assumptions, and empirically, we propose an unsupervised model selection metric ARC that aligns well with downstream accuracy. Our theory suggests an alternative understanding of contrastive learning: the role of aligning positive samples is more like a surrogate task than an ultimate goal, and the overlapped augmented views (i.e., the chaos) create a ladder for contrastive learning to gradually learn class-separated representations. The code for computing ARC is available at https://github.com/zhangq327/ARC. | Accept (Poster) |
ICLR.cc/2022/Conference | LORD: Lower-Dimensional Embedding of Log-Signature in Neural Rough Differential Equations | The problem of processing very long time-series data (e.g., a length of more than 10,000) is a long-standing research problem in machine learning. Recently, one breakthrough, called neural rough differential equations (NRDEs), has been proposed and has shown that it is able to process such data. Their main concept is to use the log-signature transform, which is known to be more efficient than the Fourier transform for irregular long time-series, to convert a very long time-series sample into a relatively shorter series of feature vectors. However, the log-signature transform causes non-trivial spatial overheads. To this end, we present the method of LOweR-Dimensional embedding of log-signature (LORD), where we define an NRDE-based autoencoder to implant the higher-depth log-signature knowledge into the lower-depth log-signature. We show that the encoder successfully combines the higher-depth and the lower-depth log-signature knowledge, which greatly stabilizes the training process and increases the model accuracy. In our experiments with benchmark datasets, the improvement ratio by our method is up to 75\% in terms of various classification and forecasting evaluation metrics. | Accept (Poster) |
ICLR.cc/2022/Conference | Knowledge Removal in Sampling-based Bayesian Inference | The right to be forgotten has been legislated in many countries, but its enforcement in the AI industry would cause unbearable costs. When single data deletion requests come, companies may need to delete the whole models learned with massive resources. Existing works propose methods to remove knowledge learned from data for explicitly parameterized models, which however are not appliable to the sampling-based Bayesian inference, {\it i.e.}, Markov chain Monte Carlo (MCMC), as MCMC can only infer implicit distributions. In this paper, we propose the first machine unlearning algorithm for MCMC. We first convert the MCMC unlearning problem into an explicit optimization problem. Based on this problem conversion, an {\it MCMC influence function} is designed to provably characterize the learned knowledge from data, which then delivers the MCMC unlearning algorithm. Theoretical analysis shows that MCMC unlearning would not compromise the generalizability of the MCMC models. Experiments on Gaussian mixture models and Bayesian neural networks confirm the effectiveness of the proposed algorithm. The code is available at \url{https://github.com/fshp971/mcmc-unlearning}. | Accept (Poster) |
ICLR.cc/2022/Conference | Cluster Tree for Nearest Neighbor Search | Tree-based algorithms are an important and widely used class of algorithms for Nearest Neighbor Search (NNS) with random partition (RP) tree being arguably the most well studied. However, in spite of possessing theoretical guarantees and strong practical performance, a major drawback of the RP tree is its lack of adaptability to the input dataset.
Inspired by recent theoretical and practical works for NNS, we attempt to remedy this by introducing ClusterTree, a new tree based algorithm. Our approach utilizes randomness as in RP trees while adapting to the underlying cluster structure of the dataset to create well-balanced and meaningful partitions. Experimental evaluations on real world datasets demonstrate improvements over RP trees and other tree based methods for NNS while maintaining efficient construction time. In addition, we show theoretically and empirically that ClusterTree finds partitions which are superior to those found by RP trees in preserving the cluster structure of the input dataset. | Desk_Rejected |
ICLR.cc/2019/Conference | Evaluating Robustness of Neural Networks with Mixed Integer Programming | Neural networks trained only to optimize for training accuracy can often be fooled by adversarial examples --- slightly perturbed inputs misclassified with high confidence. Verification of networks enables us to gauge their vulnerability to such adversarial examples. We formulate verification of piecewise-linear neural networks as a mixed integer program. On a representative task of finding minimum adversarial distortions, our verifier is two to three orders of magnitude quicker than the state-of-the-art. We achieve this computational speedup via tight formulations for non-linearities, as well as a novel presolve algorithm that makes full use of all information available. The computational speedup allows us to verify properties on convolutional and residual networks with over 100,000 ReLUs --- several orders of magnitude more than networks previously verified by any complete verifier. In particular, we determine for the first time the exact adversarial accuracy of an MNIST classifier to perturbations with bounded l-∞ norm ε=0.1: for this classifier, we find an adversarial example for 4.38% of samples, and a certificate of robustness to norm-bounded perturbations for the remainder. Across all robust training procedures and network architectures considered, and for both the MNIST and CIFAR-10 datasets, we are able to certify more samples than the state-of-the-art and find more adversarial examples than a strong first-order attack. | Accept (Poster) |
ICLR.cc/2021/Conference | Deep Learning Is Composite Kernel Learning | Recent works have connected deep learning and kernel methods. In this paper, we show that architectural choices such as convolutional layers with pooling, skip connections, make deep learning a composite kernel learning method, where the kernel is a (architecture dependent) composition of base kernels: even before training, standard deep networks have in-built structural properties that ensure their success. In particular, we build on the recently developed `neural path' framework that characterises the role of gates/masks in fully connected deep networks with ReLU activations. | Reject |
ICLR.cc/2023/Conference | A Theory of Equivalence-Preserving Program Embeddings | Program embeddings are used to solve tasks such as \textit{code clone detection} and \textit{semantic labeling}. Solutions to these \textit{semantic tasks} should be invariant to semantics-preserving program transformations. When a program embedding function satisfies this invariance, we call it an \textit{equivalence-preserving program embedding function}. We say a programming language can be \textit{tractably embedded} when we can construct an equivalence-preserving program embedding function that executes in polynomial time in program/input length and produces program embeddings that are proportional to the input length. Determining whether a programming language can be tractably embedded is the \textit{equivalence-preserving program embedding problem}. We formalize this problem and theoretically characterize when programming languages can be tractably embedded. To validate our theoretical results, we use the BERT-Tiny model to learn an equivalence-preserving program embedding function for a programming language that can be tractably embedded and show the model fails to construct an equivalence-preserving program embedding function for a similar language that is intractable to embed.
| Reject |
ICLR.cc/2023/Conference | KnowDA: All-in-One Knowledge Mixture Model for Data Augmentation in Low-Resource NLP | This paper focuses on data augmentation for low-resource NLP tasks where the training set is limited. The existing solutions either leverage task-independent heuristic rules (e.g., Synonym Replacement) or fine-tune general-purpose pre-trained language models (e.g., GPT2) using the limited training instances to produce new synthetic data. Consequently, they have trivial task-specific knowledge and are limited to yielding low-quality synthetic data. To combat this issue, we propose Knowledge Mixture Data Augmentation Model (KnowDA), a Seq2Seq language model pretrained on a mixture of diverse NLP tasks under a novel framework of Knowledge Mixture Training (KoMT). The goal of KoMT is to condense diverse NLP task-specific knowledge into the single KnowDA model
(i.e., all-in-one). The resulting KnowDA could utilize these knowledge to quickly grasp the inherent synthesis law of the target task through limited training instances. Specifically, KoMT reformulates input examples from various heterogeneous NLP tasks into a unified text-to-text format and employs denoising training objectives in different granularity to learn to reconstruct partial or complete samples. To the best of our knowledge, we are the first to attempt to apply 100+ NLP multi-task training for data augmentation. Extensive experiments show that i) the synthetic data produced by KnowDA successfully improves the performance of the strong pre-trained language
models (i.e., Bert, ALBert and Deberta) by a large margin on the low-resource NLP benchmark FewGLUE, CoNLL’03 and WikiAnn; ii) KnowDA successful transfer the task knowledge to NLP tasks whose types are seen and unseen in KoMT. | Accept: poster |
ICLR.cc/2021/Conference | Global Attention Improves Graph Networks Generalization | This paper advocates incorporating a Low-Rank Global Attention (LRGA) module, a Computation and memory efficient variant of the dot-product attention (Vaswani et al., 2017), to Graph Neural Networks (GNNs) for improving their generalization power.
To theoretically quantify the generalization properties granted by adding the LRGA module to GNNs, we focus on a specific family of expressive GNNs and show that augmenting it with LRGA provides algorithmic alignment to a powerful graph isomorphism test, namely the 2-Folklore Weisfeiler-Lehman (2-FWL) algorithm. In more detail we: (i) consider the recent Random Graph Neural Network (RGNN) (Sato et al., 2020) framework and prove that it is universal in probability; (ii) show that RGNN augmented with LRGA aligns with 2-FWL update step via polynomial kernels; and (iii) bound the sample complexity of the kernel's feature map when learned with a randomly initialized two-layer MLP.
From a practical point of view, augmenting existing GNN layers with LRGA produces state of the art results in current GNN benchmarks. Lastly, we observe that augmenting various GNN architectures with LRGA often closes the performance gap across different models. | Reject |
ICLR.cc/2023/Conference | Using semantic distance for diverse and sample efficient genetic programming | Evolutionary methods, such as genetic programming, search a space of programs to find those with good fitness, often using mutations that manipulate the syntactic structure of programs without being aware of how they affect the semantics. For applications where the semantics are highly sensitive to small syntactic mutations, or where fitness evaluation is expensive, this can make learning programs intractable.
We introduce a mutation operator that yields mutated programs that are semantically far from previously evaluated programs, while still being semantically close to their parent. For function regression, this leads to an algorithm that is one to two orders of magnitude more sample efficient than other gradient-free methods, such as genetic programming, or learning the weights of a neural network using evolutionary strategies.
We show how this method can be applied to learning architecture-specific and general purpose neural network optimizers, and to reinforcement learning loss functions. The learnt components are simple, interpretable, high performance, and contain novel features not seen before such as weight growth. | Reject |
ICLR.cc/2022/Conference | S2C2 - An orthogonal method for Semi-Supervised Learning on ambiguous labels | Semi-Supervised Learning (SSL) can decrease the required amount of labeled image data and thus the cost for deep learning. Most SSL methods assume a clear distinction between classes, but class boundaries are often ambiguous in real-world datasets due to intra- or interobserver variability. This ambiguity of annotations must be addressed as it will otherwise limit the performance of SSL and deep learning in general due to inconsistent label information. We propose SemiSupervised Classification & Clustering (S2C2) which can extend many deep SSL algorithms. S2C2 automatically estimates the ambiguity of an image and applies the respective SSL algorithm as a classification to certainly labeled data while partitioning the ambiguous data into clusters of visual similar images. We show that S2C2 results in a 7.6% better F1-score for classifications and 7.9% lower inner distance of clusters on average across multiple SSL algorithms and datasets. Moreover, the output of S2C2 can be used to decrease the ambiguity of labels with the help of human experts. Overall, a combination of Semi-Supervised Learning with our method S2C2 leads to better handling of ambiguous labels and thus realworld datasets. | Withdrawn |
ICLR.cc/2022/Conference | Discriminative Similarity for Data Clustering | Similarity-based clustering methods separate data into clusters according to the pairwise similarity between the data, and the pairwise similarity is crucial for their performance. In this paper, we propose {\em Clustering by Discriminative Similarity (CDS)}, a novel method which learns discriminative similarity for data clustering. CDS learns an unsupervised similarity-based classifier from each data partition, and searches for the optimal partition of the data by minimizing the generalization error of the learnt classifiers associated with the data partitions. By generalization analysis via Rademacher complexity, the generalization error bound for the unsupervised similarity-based classifier is expressed as the sum of discriminative similarity between the data from different classes. It is proved that the derived discriminative similarity can also be induced by the integrated squared error bound for kernel density classification. In order to evaluate the performance of the proposed discriminative similarity, we propose a new clustering method using a kernel as the similarity function, CDS via unsupervised kernel classification (CDSK), with its effectiveness demonstrated by experimental results. | Accept (Poster) |
ICLR.cc/2023/Conference | Graph schemas as abstractions for transfer learning, inference, and planning | We propose schemas as a model for abstractions that can be used for rapid transfer learning, inference, and planning. Common structured representations of concepts and behaviors---schemas---have been proposed as a powerful way to encode abstractions. Latent graph learning is emerging as a new computational model of the hippocampus to explain map learning and transitive inference. We build on this work to show that learned latent graphs in these models have a slot structure---schemas---that allow for quick knowledge transfer across environments. In a new environment, an agent can rapidly learn new bindings between the sensory stream to multiple latent schemas and select the best fitting one to guide behavior. To evaluate these graph schemas, we use two previously published challenging tasks: the memory \& planning game and one-shot StreetLearn, that are designed to test rapid task solving in novel environments. Graph schemas can be learned in far fewer episodes than previous baselines, and can model and plan in a few steps in novel variations of these tasks. We further demonstrate learning, matching, and reusing graph schemas in navigation tasks in more challenging environments with aliased observations and size variations, and show how different schemas can be composed to model larger environments. | Reject |
ICLR.cc/2022/Conference | AutoOED: Automated Optimal Experimental Design Platform with Data- and Time-Efficient Multi-Objective Optimization | We present AutoOED, an Automated Optimal Experimental Design platform powered by machine learning to accelerate discovering solutions with optimal objective trade-offs. To solve expensive multi-objective problems in a data-efficient manner, we implement popular multi-objective Bayesian optimization (MOBO) algorithms with state-of-the-art performance in a modular framework. To further accelerate the optimization in a time-efficient manner, we propose a novel strategy called Believer-Penalizer (BP), which allows batch experiments to be accelerated asynchronously without affecting performance. AutoOED serves as a testbed for machine learning researchers to quickly develop and evaluate their own MOBO algorithms. We also provide a graphical user interface (GUI) for users with little or no experience with coding, machine learning, or optimization to visualize and guide the experiment design intuitively. Finally, we demonstrate that AutoOED can control and guide real-world hardware experiments in a fully automated way without human intervention. | Reject |
ICLR.cc/2023/Conference | Computational-Unidentifiability in Representation for Fair Downstream Tasks | Deep representation learning methods are highlighted as they outperform classical algorithms in various downstream tasks, such as classification, clustering, generative models, etc. Due to their success and impact on the real world, fairness concern is rising with noticeable attention. However, the focus of the fairness problem was limited to a certain downstream task, mostly classification, and few were studied from the perspective of representation itself. We claim that the fairness problems to various downstream tasks originated from the input feature space, i.e., the learned representation space. While several studies explored fair representation for the classification task, the fair representation learning method for unsupervised learning is not actively discussed. To fill this gap, we define a new notion of fairness, computational-unidentifiability, which suggests the fairness of the representation as the distributional independence of the sensitive groups. We demonstrate motivating problems that achieving computationally-unidentifiable representation is critical for fair downstream tasks. Moreover, we propose a novel fairness metric, Fair Fréchet distance (FFD), to quantify the computational-unidentifiability and address the limitation of a well-known fairness metric for unsupervised learning, i.e., balance. The proposed metric is efficient in computation and preserves theoretical properties. We empirically validate the effectiveness of the computationally-unidentifiable representations in various downstream tasks. | Reject |
ICLR.cc/2020/Conference | Effects of Linguistic Labels on Learned Visual Representations in Convolutional Neural Networks: Labels matter! | We investigated the changes in visual representations learnt by CNNs when using different linguistic labels (e.g., trained with basic-level labels only, superordinate-level only, or both at the same time) and how they compare to human behavior when asked to select which of three images is most different. We compared CNNs with identical architecture and input, differing only in what labels were used to supervise the training. The results showed that in the absence of labels, the models learn very little categorical structure that is often assumed to be in the input. Models trained with superordinate labels (vehicle, tool, etc.) are most helpful in allowing the models to match human categorization, implying that human representations used in odd-one-out tasks are highly modulated by semantic information not obviously present in the visual input. | Reject |
ICLR.cc/2022/Conference | ANOMALY DETECTION WITH FRAME-GROUP ATTENTION IN SURVEILLANCE VIDEOS | The paper proposes an end-to-end abnormal behavior detection network to detect strenuous movements in slow moving crowds, such as running, bicycling, throwing from a height. The algorithm forms continuous video frames into a frame group and uses the frame-group feature extractor to obtain the spatio-temporal information. The implicit vector based attention mechanism will work on the extracted frame-group features to highlight the important features. We use fully connected layers to transform the space and reduce the computation. Finally, the group-pooling maps the processed frame-group features to the abnormal scores. The network input is flexible to cope with the form of video streams, and the network output is the abnormal score. The designed compound loss function will help the model improve the classification performance. This paper arranges several commonly used anomaly detection datasets and tests the algorithms on the integrated dataset. The experimental results show that the proposed algorithm has significant advantages in many objective metrics compared with other anomaly detection algorithms. | Withdrawn |
ICLR.cc/2020/Conference | NESTED LEARNING FOR MULTI-GRANULAR TASKS | Standard deep neural networks (DNNs) used for classification are trained in an end-to-end fashion for very specific tasks - object recognition, face identification, character recognition, etc. This specificity often leads to overconfident models that generalize poorly to samples that are not from the original training distribution. Moreover, they do not allow to leverage information from heterogeneously annotated data, where for example, labels may be provided with different levels of granularity. Finally, standard DNNs do not produce results with simultaneous different levels of confidence for different levels of detail, they are most commonly an all or nothing approach. To address these challenges, we introduce the problem of nested learning: how to obtain a hierarchical representation of the input such that a coarse label can be extracted first, and sequentially refine this representation to obtain successively refined predictions, all of them with the corresponding confidence. We explicitly enforce this behaviour by creating a sequence of nested information bottlenecks. Looking at the problem of nested learning from an in formation theory perspective, we design a network topology with two important properties. First, a sequence of low dimensional (nested) feature embeddings are enforced. Then we show how the explicit combination of nested outputs can improve both robustness and finer predictions. Experimental results on CIFAR-10, MNIST, and FASHION-MNIST demonstrate that nested learning outperforms the same network trained in the standard end-to-end fashion. Since the network can be naturally trained with mixed data labeled at different levels of nested details, we also study what is the most efficient way of annotating data, when a fixed training budget is given and the cost of labels increases with the levels in the nested hierarchy. | Reject |
ICLR.cc/2023/Conference | GOING BEYOND 1-WL EXPRESSIVE POWER WITH 1-LAYER GRAPH NEURAL NETWORKS | Graph neural networks have become the \textit{de facto} standard for representational learning in graphs, and have achieved SOTA in many graph-related tasks such as node classification, graph classification and link prediction. However, it has been shown that the expressive power is equivalent maximally to Weisfeiler-Lehman Test. Recently, there is a line of work aiming to enhance the expressive power of graph neural networks. In this work, we propose a more generalized variant of neural Weisfeiler-Lehman test to enhance structural representation for each node in a graph to uplift the expressive power of any graph neural network. It is shown theoretically our method is strictly more powerful than 1\&2-WL test. The Numerical experiments also demonstrate that our proposed method outperforms the standard GNNs on almost all the benchmark datasets by a large margin in most cases with significantly lower running time and memory consumption compared with other more powerful GNNs. | Reject |
ICLR.cc/2023/Conference | META-LEARNING FOR UNSUPERVISED OUTLIER DETECTION WITH OPTIMAL TRANSPORT | Automated machine learning has been widely researched and adopted in the field of supervised classification and regression, but progress in unsupervised settings has been limited. We propose a novel approach to automate outlier detection based on meta-learning from previous datasets with outliers. Our premise is that the selection of the optimal outlier detection technique depends on inherent properties of the data distribution. We leverage the Gromov-Wasserstein distance in particular, to find the dataset with the most similar underlying distribution, and then apply the outlier detection techniques that proved to work best for that data distribution. We evaluate the robustness of our approach and find that it outperforms the state of the art methods in unsupervised outlier detection. This approach can also be easily generalized to automate other unsupervised settings. | Withdrawn |
ICLR.cc/2022/Conference | Node-Level Differentially Private Graph Neural Networks | Graph neural networks (GNNs) are a popular technique for modelling graph-structured data that compute node-level predictions via aggregation of information from the local neighborhood of each node. However, this aggregation implies increased risk of revealing sensitive information, as a node can participate in the inference for multiple nodes. This implies that standard privacy preserving machine learning techniques like differentially private stochastic gradient descent (DP-SGD) – which are designed for situations where each node/data point participate in inference of one point only – either do not apply or lead to inaccurate solutions. In this work, we formally define the problem of learning 1-layer GNNs with node-level privacy, and provide a method for the problem with a strong differential privacy guarantee. Even though each node can be involved in the inference for multiple nodes, by employing a careful sensitivity analysis and a non-trivial extension of the privacy-by-amplification technique, our method is able to provide accurate solutions with solid privacy parameters. Empirical evaluation on standard benchmarks demonstrates that our method is indeed able to learn accurate privacy preserving GNNs, while still outperforming standard non-private methods that completely ignore graph information. | Reject |
ICLR.cc/2019/Conference | IEA: Inner Ensemble Average within a convolutional neural network | Ensemble learning is a method of combining multiple trained models to improve model accuracy. We propose the usage of such methods, specifically ensemble average, inside Convolutional Neural Network (CNN) architectures by replacing the single convolutional layers with Inner Average Ensembles (IEA) of multiple convolutional layers. Empirical results on different benchmarking datasets show that CNN models using IEA outperform those with regular convolutional layers and advances the state of art. A visual and a similarity score analysis of the features generated from IEA explains why it boosts the model performance. | Reject |
ICLR.cc/2019/Conference | A Model Cortical Network for Spatiotemporal Sequence Learning and Prediction | In this paper we developed a hierarchical network model, called Hierarchical Prediction Network (HPNet) to understand how spatiotemporal memories might be learned and encoded in a representational hierarchy for predicting future video frames. The model is inspired by the feedforward, feedback and lateral recurrent circuits in the mammalian hierarchical visual system. It assumes that spatiotemporal memories are encoded in the recurrent connections within each level and between different levels of the hierarchy. The model contains a feed-forward path that computes and encodes spatiotemporal features of successive complexity and a feedback path that projects interpretation from a higher level to the level below. Within each level, the feed-forward path and the feedback path intersect in a recurrent gated circuit that integrates their signals as well as the circuit's internal memory states to generate a prediction of the incoming signals. The network learns by comparing the incoming signals with its prediction, updating its internal model of the world by minimizing the prediction errors at each level of the hierarchy in the style of {\em predictive self-supervised learning}. The network processes data in blocks of video frames rather than a frame-to-frame basis. This allows it to learn relationships among movement patterns, yielding state-of-the-art performance in long range video sequence predictions in benchmark datasets. We observed that hierarchical interaction in the network introduces sensitivity to memories of global movement patterns even in the population representation of the units in the earliest level. Finally, we provided neurophysiological evidence, showing that neurons in the early visual cortex of awake monkeys exhibit very similar sensitivity and behaviors. These findings suggest that predictive self-supervised learning might be an important principle for representational learning in the visual cortex. | Reject |
ICLR.cc/2022/Conference | Structured Stochastic Gradient MCMC | Stochastic gradient Markov Chain Monte Carlo (SGMCMC) is considered the gold standard for Bayesian inference in large-scale models, such as Bayesian neural networks. Since practitioners face speed versus accuracy tradeoffs in these models, variational inference (VI) is often the preferable option. Unfortunately, VI makes strong assumptions on both the factorization and functional form of the posterior. In this work, we propose a new non-parametric variational approximation that makes no assumptions about the approximate posterior's functional form and allows practitioners to specify the exact dependencies the algorithm should respect or break. The approach relies on a new Langevin-type algorithm that operates on a modified energy function, where parts of the latent variables are averaged over samples from earlier iterations of the Markov chain. This way, statistical dependencies can be broken in a controlled way, allowing the chain to mix faster. This scheme can be further modified in a ``dropout'' manner, leading to even more scalability. By implementing the scheme on a ResNet-20 architecture, we obtain better predictive likelihoods and faster mixing time than full SGMCMC. | Reject |
ICLR.cc/2022/Conference | Learning Audio-Visual Dereverberation | Reverberation from audio reflecting off surfaces and objects in the environment not only degrades the quality of speech for human perception, but also severely impacts the accuracy of automatic speech recognition. Prior work attempts to remove reverberation based on the audio modality only. Our idea is to learn to dereverberate speech from audio-visual observations. The visual environment surrounding a human speaker reveals important cues about the room geometry, materials, and speaker location, all of which influence the precise reverberation effects in the audio stream. We introduce Visually-Informed Dereverberation of Audio (VIDA), an end-to-end approach that learns to remove reverberation based on both the observed sounds and visual scene. In support of this new task, we develop a large-scale dataset that uses realistic acoustic renderings of speech in real-world 3D scans of homes offering a variety of room acoustics. Demonstrating our approach on both simulated and real imagery for speech enhancement, speech recognition, and speaker identification, we show it achieves state-of-the-art performance and substantially improves over traditional audio-only methods. | Reject |
ICLR.cc/2022/Conference | Adaptive Learning of Tensor Network Structures | Tensor Networks (TN) offer a powerful framework to efficiently represent very high-dimensional objects. TN have recently shown their potential for machine learning applications and offer a unifying view of common tensor decomposition models such as Tucker, tensor train (TT) and tensor ring (TR). However, identifying the best tensor network structure from data for a given task is challenging. In this work, we leverage the TN formalism to develop a generic and efficient adaptive algorithm to jointly learn the structure and the parameters of a TN from data. Our method is based on a simple greedy approach starting from a rank one tensor and successively identifying the most promising tensor network edges for small rank increments. Our algorithm can adaptively identify TN structures with small number of parameters that effectively optimize any differentiable objective function. Experiments on tensor decomposition, tensor completion and model compression tasks demonstrate the effectiveness of the proposed algorithm. In particular, our method outperforms the state-of-the-art evolutionary topology search [Li and Sun, 2020] for tensor decomposition of images (while being orders of magnitude faster) and finds efficient tensor network structures to compress neural networks outperforming popular TT based approaches [Novikov et al., 2015]. | Reject |
ICLR.cc/2023/Conference | Hybrid RL: Using both offline and online data can make RL efficient | We consider a hybrid reinforcement learning setting (Hybrid RL), in which an agent has access to an offline dataset and the ability to collect experience via real-world online interaction. The framework mitigates the challenges that arise in both pure offline and online RL settings, allowing for the design of simple and highly effective algorithms, in both theory and practice. We demonstrate these advantages by adapting the classical Q learning/iteration algorithm to the hybrid setting, which we call Hybrid Q-Learning or Hy-Q. In our theoretical results, we prove that the algorithm is both computationally and statistically efficient whenever the offline dataset supports a high-quality policy and the environment has bounded bilinear rank. Notably, we require no assumptions on the coverage provided by the initial distribution, in contrast with guarantees for policy gradient/iteration methods. In our experimental results, we show that Hy-Q with neural network function approximation outperforms state-of-the-art online, offline, and hybrid RL baselines on challenging benchmarks, including Montezuma’s Revenge. | Accept: poster |
ICLR.cc/2019/Conference | Directed-Info GAIL: Learning Hierarchical Policies from Unsegmented Demonstrations using Directed Information | The use of imitation learning to learn a single policy for a complex task that has multiple modes or hierarchical structure can be challenging. In fact, previous work has shown that when the modes are known, learning separate policies for each mode or sub-task can greatly improve the performance of imitation learning. In this work, we discover the interaction between sub-tasks from their resulting state-action trajectory sequences using a directed graphical model. We propose a new algorithm based on the generative adversarial imitation learning framework which automatically learns sub-task policies from unsegmented demonstrations. Our approach maximizes the directed information flow in the graphical model between sub-task latent variables and their generated trajectories. We also show how our approach connects with the existing Options framework, which is commonly used to learn hierarchical policies. | Accept (Poster) |
ICLR.cc/2023/Conference | Token Turing Machines | We propose Token Turing Machines (TTM), a sequential, autoregressive Transformer model with memory for real-world sequential decision making. Our model is inspired by the seminal Neural Turing Machine, and has an external memory consisting of a set of tokens which summarise the previous history. This memory is efficiently addressed, read and written using a Transformer as the processing unit/controller at each step. The model's memory module ensures that a new observation will only be processed with the contents of the memory (and not the entire history), meaning that it can efficiently process long sequences with a bounded computational cost at each step. We show that TTM outperforms other alternatives, such as other Transformer models designed for long sequences and recurrent neural networks, on two real-world sequential decision making tasks: online temporal activity localization from videos and vision-based robot action policy learning. | Withdrawn |
ICLR.cc/2023/Conference | Preserving In-Context Learning Ability in Large Language Model Fine-tuning | Pretrained large language models (LLMs) are strong in-context learners that are able to perform few-shot learning without changing model parameters. However, as we show, fine-tuning an LLM on any specific task generally destroys its in-context ability. We discover an important cause of this loss, format specialization, where the model overfits to the format of the fine-tuned task and is unable to output anything beyond this format. We further show that format specialization happens at the beginning of fine-tuning. To solve this problem, we propose Prompt Tuning with MOdel Tuning (ProMoT), a simple yet effective two-stage fine-tuning framework that preserves in-context abilities of the pretrained model substantially better than vanilla fine-tuning. ProMoT first trains a soft prompt for the fine-tuning target task, and then fine-tunes the model itself with this soft prompt attached. ProMoT offloads task-specific formats into the soft prompt that can be easily removed when doing other in-context tasks. We fine-tune mT5 XXL with ProMoT on natural language inference (NLI) and English-French translation and evaluate the in-context abilities of the resulting models on 8 different NLP datasets including classification, summarization, translation and question answering. ProMoT achieves similar performance on the fine-tuned tasks compared with vanilla fine-tuning, but with much less reduction of in-context learning performances across the board. More importantly, ProMoT shows remarkable generalization ability on tasks that have different formats, e.g. fine-tuning on a NLI binary classification task improves the model's in-context ability to do summarization (+0.53 Rouge-2 score compared to the pretrained model), making ProMoT a promising method to build general purpose capabilities such as grounding and reasoning into LLMs with small but high quality datasets. | Withdrawn |
ICLR.cc/2022/Conference | Fingerprinting Multi-exit Deep Neural Network Models via Inference Time | Transforming large deep neural network (DNN) models into the multi-exit architectures can overcome the overthinking issue and distribute a large DNN model on resource-constrained scenarios (e.g. IoT frontend devices and backend servers) for inference and transmission efficiency. Nevertheless, intellectual property (IP) protection for the multi-exit models in the wild is still an unsolved challenge. Previous efforts to verify DNN model ownership mainly rely on querying the model with specific samples and checking the responses, e.g., DNN watermarking and fingerprinting. However, they are vulnerable to adversarial settings such as adversarial training and are not suitable for the IP verification for multi-exit DNN models. In this paper, we propose a novel approach to fingerprint multi-exit models via inference time rather than inference predictions. Specifically, we design an effective method to generate a set of fingerprint samples to craft the inference process with a unique and robust inference time cost as the evidence for model ownership. We conduct extensive experiments to prove the uniqueness and robustness of our method on three structures (ResNet-56, VGG-16, and MobileNet) and three datasets (CIFAR-10, CIFAR-100, and Tiny-ImageNet) under comprehensive adversarial settings. | Withdrawn |
ICLR.cc/2023/Conference | Diversify and Disambiguate: Out-of-Distribution Robustness via Disagreement | Real-world machine learning problems often exhibit shifts between the source and target distributions, in which source data does not fully convey the desired behavior on target inputs. Different functions that achieve near-perfect source accuracy can make differing predictions on test inputs, and such ambiguity makes robustness to distribution shifts challenging. We propose DivDis, a simple two-stage framework for identifying and resolving ambiguity in data. DivDis first learns a diverse set of hypotheses that achieve low source loss but make differing predictions on target inputs. We then disambiguate by selecting one of the discovered functions using additional information, for example, a small number of target labels. Our experimental evaluation shows improved performance in subpopulation shift and domain generalization settings, demonstrating that DivDis can scalably adapt to distribution shifts in image and text classification benchmarks. | Accept: poster |
ICLR.cc/2022/Conference | Adjoined Networks: A Training Paradigm with Applications to Network Compression | Compressing deep neural networks while maintaining accuracy is important when we want to deploy large, powerful models in production and/or edge devices. One common technique used to achieve this goal is knowledge distillation. Typically, the output of a static pre-defined teacher (a large base network) is used as soft labels to train and transfer information to a student (or smaller) network. In this paper, we introduce Adjoined Networks, or AN, a learning paradigm that trains both the original base network and the smaller compressed network together. In our training approach, the parameters of the smaller network are shared across both the base and the compressed networks. Using our training paradigm, we can simultaneously compress (the student network) and regularize (the teacher network) any architecture. In this paper, we focus on popular CNN-based architectures used for computer vision tasks. We conduct an extensive experimental evaluation of our training paradigm on various large-scale datasets. Using ResNet-50 as the base network, AN achieves 71.8% top-1 accuracy with only 1.8M parameters and 1.6 GFLOPs on the ImageNet data-set. We further propose Differentiable Adjoined Networks (DAN), a training paradigm that augments AN by using neural architecture search to jointly learn both the width and the weights for each layer of the smaller network. DAN achieves ResNet-50 level accuracy on ImageNet with $3.8\times$ fewer parameters and $2.2\times$ fewer FLOPs. | Withdrawn |
ICLR.cc/2021/Conference | Data Transformer for Anomalous Trajectory Detection | Anomaly detection is an important task in many traffic applications. Methods based on convolutional neural networks reach state-of-the-art accuracy; however, they typically rely on supervised training with large labeled data and the trained network is only applicable to the intersection that the training data are collected from. Considering that anomaly data are generally hard to obtain, we present data transformation methods for converting data obtained from one intersection to other intersections to mitigate the effort of training data collection. We demonstrate our methods on the task of anomalous trajectory detection and leverage an unsupervised method that require only normal trajectories for network training. We proposed a general model and a universal model for our transformation methods. The general model focuses on saving data collection effort; while the universal model aims at training a universal network for being used by other intersections. We evaluated our methods on the dataset with trajectories collected from GTA V virtual world. The experimental results show that with significant reduction in data collecting and network training efforts, our methods still can achieve state-of-the-art accuracy for anomalous trajectory detection. | Desk_Rejected |
ICLR.cc/2021/Conference | Non-Asymptotic PAC-Bayes Bounds on Generalisation Error | Constructing non-vacuous PAC-Bayes bounds on generalization errors for un- bounded risk functionals, especially in the non-asymptotic regime, is an active area of research. However, current state of the art results are applicable only in some very specialized cases. In this work, we give an integrability condition which exactly characterizes when any risk functional, for a given data set and model space, admits such bounds using the Levy-Khintchine theorem. Further, we de- rive a Bahadur-Rao type exact asymptotic bound, which is much sharper than a traditional Chernoff type inequality, especially in the under-sampled regime. These bounds give us the flexibility to construct data or model-dependent consistency promoting updates to a data-free prior, which provably improves the generalization performance. | Withdrawn |
ICLR.cc/2020/Conference | Better Knowledge Retention through Metric Learning | In a continual learning setting, new categories may be introduced over time, and an ideal learning system should perform well on both the original categories and the new categories. While deep neural nets have achieved resounding success in the classical setting, they are known to forget about knowledge acquired in prior episodes of learning if the examples encountered in the current episode of learning are drastically different from those encountered in prior episodes. This makes deep neural nets ill-suited to continual learning. In this paper, we propose a new model that can both leverage the expressive power of deep neural nets and is resilient to forgetting when new categories are introduced. We demonstrate an improvement in terms of accuracy on original classes compared to a vanilla deep neural net. | Reject |
ICLR.cc/2020/Conference | BERTScore: Evaluating Text Generation with BERT | We propose BERTScore, an automatic evaluation metric for text generation. Analogously to common metrics, BERTScore computes a similarity score for each token in the candidate sentence with each token in the reference sentence. However, instead of exact matches, we compute token similarity using contextual embeddings. We evaluate using the outputs of 363 machine translation and image captioning systems. BERTScore correlates better with human judgments and provides stronger model selection performance than existing metrics. Finally, we use an adversarial paraphrase detection task and show that BERTScore is more robust to challenging examples compared to existing metrics. | Accept (Poster) |
ICLR.cc/2022/Conference | Red Alarm for Pre-trained Models: Universal Vulnerability to Neuron-Level Backdoor Attacks | The pre-training-then-fine-tuning paradigm has been widely used in deep learning. Due to the huge computation cost for pre-training, practitioners usually download pre-trained models from the Internet and fine-tune them on downstream datasets while the downloaded models may suffer backdoor attacks. Different from previous attacks aiming at a target task, we show that a backdoored pre-trained model can behave maliciously in various downstream tasks without foreknowing task information. Attackers can restrict the output representations of trigger-embedded samples to arbitrary predefined values through additional training, namely Neuron-level Backdoor Attack (NeuBA). Since fine-tuning has little effect on model parameters, the fine-tuned model will retain the backdoor functionality and predict a specific label for the samples embedded with the same trigger. To provoke multiple labels in a specific task, attackers can introduce several triggers with contrastive predefined values. In the experiments of both natural language processing (NLP) and computer vision (CV), we show that NeuBA can well control the predictions for trigger-embedded instances with different trigger designs. Our findings sound a red alarm for the wide use of pre-trained models. Finally, we apply several defense methods to NeuBA and find that model pruning is a promising technique to resist NeuBA by omitting backdoored neurons. | Reject |
ICLR.cc/2023/Conference | Toward Adversarial Training on Contextualized Language Representation | Beyond the success story of adversarial training (AT) in the recent text domain on top of pre-trained language models (PLMs), our empirical study showcases the inconsistent gains from AT on some tasks, e.g. commonsense reasoning, named entity recognition. This paper investigates AT from the perspective of the contextualized language representation outputted by PLM encoders. We find the current AT attacks lean to generate sub-optimal adversarial examples that can fool the decoder part but have a minor effect on the encoder. However, we find it necessary to effectively deviate the latter one to allow AT to gain. Based on the observation, we propose simple yet effective \textit{Contextualized representation-Adversarial Training} (CreAT), in which the attack is explicitly optimized to deviate the contextualized representation of the encoder. It allows a global optimization of adversarial examples that can fool the entire model. We also find CreAT gives rise to a better direction to optimize the adversarial examples, to let them less sensitive to hyperparameters. Compared to AT, CreAT produces consistent performance gains on a wider range of tasks and is proven to be more effective for language pre-training where only the encoder part is kept for downstream tasks. We achieve the new state-of-the-art performances on a series of challenging benchmarks, e.g. AdvGLUE (59.1 $ \rightarrow $ 61.1), HellaSWAG (93.0 $ \rightarrow $ 94.9), ANLI (68.1 $ \rightarrow $ 69.3). | Accept: poster |
ICLR.cc/2021/Conference | Generalized Multimodal ELBO | Multiple data types naturally co-occur when describing real-world phenomena and learning from them is a long-standing goal in machine learning research. However, existing self-supervised generative models approximating an ELBO are not able to fulfill all desired requirements of multimodal models: their posterior approximation functions lead to a trade-off between the semantic coherence and the ability to learn the joint data distribution. We propose a new, generalized ELBO formulation for multimodal data that overcomes these limitations. The new objective encompasses two previous methods as special cases and combines their benefits without compromises. In extensive experiments, we demonstrate the advantage of the proposed method compared to state-of-the-art models in self-supervised, generative learning tasks. | Accept (Poster) |
ICLR.cc/2022/Conference | Lossless Compression with Probabilistic Circuits | Despite extensive progress on image generation, common deep generative model architectures are not easily applied to lossless compression. For example, VAEs suffer from a compression cost overhead due to their latent variables. This overhead can only be partially eliminated with elaborate schemes such as bits-back coding, often resulting in poor single-sample compression rates. To overcome such problems, we establish a new class of tractable lossless compression models that permit efficient encoding and decoding: Probabilistic Circuits (PCs). These are a class of neural networks involving $|p|$ computational units that support efficient marginalization over arbitrary subsets of the $D$ feature dimensions, enabling efficient arithmetic coding. We derive efficient encoding and decoding schemes that both have time complexity $\mathcal{O} (\log(D) \cdot |p|)$, where a naive scheme would have linear costs in $D$ and $|p|$, making the approach highly scalable. Empirically, our PC-based (de)compression algorithm runs 5-40 times faster than neural compression algorithms that achieve similar bitrates. By scaling up the traditional PC structure learning pipeline, we achieve state-of-the-art results on image datasets such as MNIST. Furthermore, PCs can be naturally integrated with existing neural compression algorithms to improve the performance of these base models on natural image datasets. Our results highlight the potential impact that non-standard learning architectures may have on neural data compression. | Accept (Spotlight) |
ICLR.cc/2022/Conference | Gradient Imbalance and solution in Online Continual learning | Most existing techniques for online continual learning are based on experience-replay. In this approach, a memory buffer is used to save some data from past tasks for dealing with catastrophic forgetting. In training, a small batch of data from the data stream of the current task and some sampled data from a memory buffer are used jointly to update or train the current model. In this paper, we study the experience replay-based approach from a new angle, gradient imbalance. We first investigate and analyze this phenomenon experimentally from two perspectives: imbalance of samples introduced by experience replay and sequence of classes introduced by incremental learning. To our knowledge, this problem has not been studied before and it significantly limits the performance of online continual learning. Based on observations from experiments and theoretical analysis, a new learning strategy and a new loss are proposed to deal with the problem. Empirical evaluation shows that GAD helps improve the online CL performance by more than 11% in accuracy. | Withdrawn |
ICLR.cc/2018/Conference | Coupled Ensembles of Neural Networks | We investigate in this paper the architecture of deep convolutional networks. Building on existing state of the art models, we propose a reconfiguration of the model parameters into several parallel branches at the global network level, with each branch being a standalone CNN. We show that this arrangement is an efficient way to significantly reduce the number of parameters while at the same time improving the performance. The use of branches brings an additional form of regularization. In addition to splitting the parameters into parallel branches, we propose a tighter coupling of these branches by averaging their log-probabilities. The tighter coupling favours the learning of better representations, even at the level of the individual branches, as compared to when each branch is trained independently. We refer to this branched architecture as "coupled ensembles". The approach is very generic and can be applied with almost any neural network architecture. With coupled ensembles of DenseNet-BC and parameter budget of 25M, we obtain error rates of 2.92%, 15.68% and 1.50% respectively on CIFAR-10, CIFAR-100 and SVHN tasks. For the same parameter budget, DenseNet-BC has an error rate of 3.46%, 17.18%, and 1.8% respectively. With ensembles of coupled ensembles, of DenseNet-BC networks, with 50M total parameters, we obtain error rates of 2.72%, 15.13% and 1.42% respectively on these tasks. | Reject |
ICLR.cc/2023/Conference | Wasserstein Barycenter-based Model Fusion and Linear Mode Connectivity of Neural Networks | Based on the concepts of Wasserstein barycenter (WB) and Gromov-Wasserstein barycenter (GWB), we propose a unified mathematical framework for neural network (NN) model fusion and utilize it to reveal new insights about the linear mode connectivity of SGD solutions. In our framework, the fusion occurs in a layer-wise manner and builds on an interpretation of a node in a network as a function of the layer preceding it.
The versatility of our mathematical framework allows us to talk about model fusion and linear mode connectivity for a broad class of NNs, including fully connected NN, CNN, ResNet, RNN, and LSTM, in each case exploiting the specific structure of the network architecture. We present extensive numerical experiments to: 1) illustrate the strengths of our approach in relation to other model fusion methodologies and 2) from a certain perspective, provide new empirical evidence for recent conjectures which say that two local minima found by gradient-based methods end up lying on the same basin of the loss landscape after a proper permutation of weights is applied to one of the models. | Reject |
ICLR.cc/2023/Conference | Causal Representation Learning for Instantaneous and Temporal Effects in Interactive Systems | Causal representation learning is the task of identifying the underlying causal variables and their relations from high-dimensional observations, such as images. Recent work has shown that one can reconstruct the causal variables from temporal sequences of observations under the assumption that there are no instantaneous causal relations between them. In practical applications, however, our measurement or frame rate might be slower than many of the causal effects. This effectively creates ``instantaneous'' effects and invalidates previous identifiability results. To address this issue, we propose iCITRIS, a causal representation learning method that allows for instantaneous effects in intervened temporal sequences when intervention targets can be observed, e.g., as actions of an agent. iCITRIS identifies the potentially multidimensional causal variables from temporal observations, while simultaneously using a differentiable causal discovery method to learn their causal graph. In experiments on three datasets of interactive systems, iCITRIS accurately identifies the causal variables and their causal graph. | Accept: poster |
ICLR.cc/2022/Conference | Chemical-Reaction-Aware Molecule Representation Learning | Molecule representation learning (MRL) methods aim to embed molecules into a real vector space. However, existing SMILES-based (Simplified Molecular-Input Line-Entry System) or GNN-based (Graph Neural Networks) MRL methods either take SMILES strings as input that have difficulty in encoding molecule structure information, or over-emphasize the importance of GNN architectures but neglect their generalization ability. Here we propose using chemical reactions to assist learning molecule representation. The key idea of our approach is to preserve the equivalence of molecules with respect to chemical reactions in the embedding space, i.e., forcing the sum of reactant embeddings and the sum of product embeddings to be equal for each chemical equation. This constraint is proven effective to 1) keep the embedding space well-organized and 2) improve the generalization ability of molecule embeddings. Moreover, our model can use any GNN as the molecule encoder and is thus agnostic to GNN architectures. Experimental results demonstrate that our method achieves state-of-the-art performance in a variety of downstream tasks, e.g., reaction product prediction, molecule property prediction, reaction classification, and graph-edit-distance prediction. The code is available at https://github.com/hwwang55/MolR. | Accept (Poster) |
ICLR.cc/2019/Conference | MLPrune: Multi-Layer Pruning for Automated Neural Network Compression | Model compression can significantly reduce the computation and memory footprint of large neural networks. To achieve a good trade-off between model size and accuracy, popular compression techniques usually rely on hand-crafted heuristics and
require manually setting the compression ratio of each layer. This process is typically costly and suboptimal. In this paper, we propose a Multi-Layer Pruning method (MLPrune), which is theoretically sound, and can automatically decide appropriate compression ratios for all layers. Towards this goal, we use an efficient approximation of the Hessian as our pruning criterion, based on a Kronecker-factored Approximate Curvature method. We demonstrate the effectiveness of our method on several datasets and architectures, outperforming previous state-of-the-art by a large margin. Our experiments show that we can compress AlexNet and VGG16 by 25x without loss in accuracy on ImageNet. Furthermore, our method has much fewer hyper-parameters and requires no expert knowledge. | Reject |
ICLR.cc/2022/Conference | Explore and Control with Adversarial Surprise | Unsupervised reinforcement learning (RL) studies how to leverage environment statistics to learn useful behaviors without the cost of reward engineering. However, a central challenge in unsupervised RL is to extract behaviors that meaningfully affect the world and cover the range of possible outcomes, without getting distracted by inherently unpredictable, uncontrollable, and stochastic elements in the environment. To this end, we propose an unsupervised RL method designed for high-dimensional, stochastic environments based on an adversarial game between two policies (which we call Explore and Control) controlling a single body and competing over the amount of observation entropy the agent experiences. The Explore agent seeks out states that maximally surprise the Control agent, which in turn aims to minimize surprise, and thereby manipulate the environment to return to familiar and predictable states. The competition between these two policies drives them to seek out increasingly surprising parts of the environment while learning to gain mastery over them. We show formally that the resulting algorithm maximizes coverage of the underlying state in block MDPs with stochastic observations, providing theoretical backing to our hypothesis that this procedure avoids uncontrollable and stochastic distractions. Our experiments further demonstrate that Adversarial Surprise leads to the emergence of complex and meaningful skills, and outperforms state-of-the-art unsupervised reinforcement learning methods in terms of both exploration and zero-shot transfer to downstream tasks. | Reject |
ICLR.cc/2021/Conference | A Unified View on Graph Neural Networks as Graph Signal Denoising | Graph Neural Networks (GNNs) have risen to prominence in learning representations for graph structured data. A single GNN layer typically consists of a feature transformation and a feature aggregation operation. The former normally uses feed-forward networks to transform features, while the latter aggregates the transformed features over the graph. Numerous recent works have proposed GNN models with different designs in the aggregation operation. In this work, we establish mathematically that the aggregation processes in a group of representative GNN models including GCN, GAT, PPNP, and APPNP can be regarded as (approximately) solving a graph denoising problem with a smoothness assumption. Such a unified view across GNNs not only provides a new perspective to understand a variety of aggregation operations but also enables us to develop a unified graph neural network framework UGNN. To demonstrate its promising potential, we instantiate a novel GNN model, ADA-UGNN, derived from UGNN, to handle graphs with adaptive smoothness across nodes. Comprehensive experiments show the effectiveness of ADA-UGNN. | Reject |
ICLR.cc/2020/Conference | Teacher-Student Compression with Generative Adversarial Networks | More accurate machine learning models often demand more computation and memory at test time, making them difficult to deploy on CPU- or memory-constrained devices. Teacher-student compression (TSC), also known as distillation, alleviates this burden by training a less expensive student model to mimic the expensive teacher model while maintaining most of the original accuracy. However, when fresh data is unavailable for the compression task, the teacher's training data is typically reused, leading to suboptimal compression. In this work, we propose to augment the compression dataset with synthetic data from a generative adversarial network (GAN) designed to approximate the training data distribution. Our GAN-assisted TSC (GAN-TSC) significantly improves student accuracy for expensive models such as large random forests and deep neural networks on both tabular and image datasets. Building on these results, we propose a comprehensive metric—the TSC Score—to evaluate the quality of synthetic datasets based on their induced TSC performance. The TSC Score captures both data diversity and class affinity, and we illustrate its benefits over the popular Inception Score in the context of image classification. | Reject |
ICLR.cc/2022/Conference | Exploring and Evaluating Personalized Models for Code Generation | Large Transformer models achieved the state-of-the-art status for Natural Language Understanding and are increasingly the baseline architecture for source code generation models. Transformers are usually pre-trained on a large unsupervised corpus, learning token representations and transformations relevant to modeling generally available text, and then fine-tuned on a particular task of interest. While fine-tuning is a tried-and-true method for adapting a model to a new domain, for example question-answering on a given topic or a source code generation model, generalization remains an on-going challenge. Here we explore the ability of various levels of model fine-tuning to improve generalization by personalized fine-tuning. In the context of generating unit tests for Java methods, here we evaluate learning to personalize to a specific project using several methods to personalize transformer models for unit test generation for a specific Java project. We consider three fine-tuning approaches: (i) custom fine-tuning, which allows all the model parameters to be tuned; (ii) lightweight fine-tuning, which freezes most of the model's parameters, allowing a tuning of the token embeddings and softmax layer or the final layer alone; (iii) prefix tuning, which keeps language model parameters frozen, but optimizes a small project-specific prefix vector. Each of these techniques offers a different trade-off in total compute cost and prediction performance, which we evaluate by code and task-specific metrics, training time, and total computational operations. We compare these fine-tuning strategies for code generation and discuss the potential generalization and cost benefits of each in deployment scenarios. | Reject |
ICLR.cc/2023/Conference | Towards Online Real-Time Memory-based Video Inpainting Transformers | Video inpainting tasks have seen significant improvements in the past years with the rise of deep neural networks and, in particular, vision transformers. Although these models show promising reconstruction quality and temporal consistency, they are still unsuitable for live videos, one of the last steps to make them completely convincing and usable. The main limitations are that these state-of-the-art models inpaint using the whole video (offline processing) and show an insufficient frame rate. In our approach, we propose a framework to adapt existing inpainting transformers to these constraints by memorizing and refining redundant computations while maintaining a decent inpainting quality. Using this framework with some of the most recent inpainting models, we show great online results with a consistent throughput above 20 frames per second. Code and pretrained models will be made available upon acceptance. | Reject |
ICLR.cc/2021/Conference | Improving Zero-Shot Neural Architecture Search with Parameters Scoring | The exceptional success of deep learning comes at the cost of long training sessions, and a slow iterative process of proposing new architectures that have to be hand-engineered through years of experience. Neural Architecture Search (NAS) is the line of research that tries to automatically design architectures with better performances at a given task. The performance of a network in a task can be predicted by a score, even before the network is trained: this is referred to as zero-shot NAS. However, the existing score remains unreliable for architectures with high accuracy. We develop in this direction by exploring different related scores. We study their time efficiency and we improve on their dependence with the final accuracy, especially for high values of the score. We propose a monotonicity metric to evaluate the adequate relative scoring of the architectures, as a way to avoid imposing a linearity assumption too early. We find that our use of noise improves the score, but a more substantial improvement comes when the evaluation of the score is done in the parameter space. We hope this effort will help clarify promising directions to speed up automatic discovery of good neural architectures without training. | Withdrawn |
ICLR.cc/2022/Conference | For Manifold Learning, Deep Neural Networks Can be Locality Sensitive Hash Functions | It is well established that training deep neural networks gives useful representations that capture essential features of the inputs. However, these representations are poorly understood in theory and practice. In the context of supervised learning an important question is whether these representations capture features informative for classification, while filtering out non-informative noisy ones. We present a formal framework to study this question by considering a generative process where each class is associated with a high-dimensional manifold and different classes define different manifolds. Under this model, each input is produced using two latent vectors: (i) a ``manifold identifier" $\gamma$ and; (ii)~a ``transformation parameter" $\theta$ that shifts examples along the surface of a manifold. E.g., $\gamma$ might represent a canonical image of a dog, and $\theta$ might stand for variations in pose, background or lighting. We provide theoretical evidence that neural representations can be viewed as LSH-like functions that map each input to an embedding that is a function of solely the informative $\gamma$ and invariant to $\theta$, effectively recovering the manifold identifier . We formally show that we get one-shot learning to unseen classes as an important consequence of this behavior. | Withdrawn |
ICLR.cc/2020/Conference | Mogrifier LSTM | Many advances in Natural Language Processing have been based upon more expressive models for how inputs interact with the context in which they occur. Recurrent networks, which have enjoyed a modicum of success, still lack the generalization and systematicity ultimately required for modelling language. In this work, we propose an extension to the venerable Long Short-Term Memory in the form of mutual gating of the current input and the previous output. This mechanism affords the modelling of a richer space of interactions between inputs and their context. Equivalently, our model can be viewed as making the transition function given by the LSTM context-dependent. Experiments demonstrate markedly improved generalization on language modelling in the range of 3–4 perplexity points on Penn Treebank and Wikitext-2, and 0.01–0.05 bpc on four character-based datasets. We establish a new state of the art on all datasets with the exception of Enwik8, where we close a large gap between the LSTM and Transformer models.
| Accept (Talk) |
ICLR.cc/2020/Conference | Aging Memories Generate More Fluent Dialogue Responses with Memory Networks | The integration of a Knowledge Base (KB) into a neural dialogue agent is one of the key challenges in Conversational AI. Memory networks has proven to be effective to encode KB information into an external memory to thus generate more fluent and informed responses. Unfortunately, such memory becomes full of latent representations during training, so the most common strategy is to overwrite old memory entries randomly.
In this paper, we question this approach and provide experimental evidence showing that conventional memory networks generate many redundant latent vectors resulting in overfitting and the need for larger memories. We introduce memory dropout as an automatic technique that encourages diversity in the latent space by 1) Aging redundant memories to increase their probability of being overwritten during training 2) Sampling new memories that summarize the knowledge acquired by redundant memories. This technique allows us to incorporate Knowledge Bases to achieve state-of-the-art dialogue generation in the Stanford Multi-Turn Dialogue dataset. Considering the same architecture, its use provides an improvement of +2.2 BLEU points for the automatic generation of responses and an increase of +8.1% in the recognition of named entities. | Withdrawn |
ICLR.cc/2023/Conference | Continuously Parameterized Mixture Models | Mixture models are universal approximators of smooth densities but are difficult to utilize in complicated datasets due to restrictions on typically available modes and challenges with initialiations.
We show that by continuously parameterizing a mixture of factor analyzers using a learned ordinary differential equation, we can improve the fit of mixture models over direct methods.
Once trained, the mixture components can be extracted and the neural ODE can be discarded, leaving us with an effective, but low-resource model.
We additionally explore the use of a training curriculum from an easy-to-model latent space extracted from a normalizing flow to the more complex input space and show that the smooth curriculum helps to stabilize and improve results with and without the continuous parameterization.
Finally, we introduce a hierarchical version of the model to enable more flexible, robust classification and clustering, and show substantial improvements against traditional parameterizations of GMMs. | Withdrawn |
ICLR.cc/2023/Conference | The ethical ambiguity of AI data enrichment: Measuring gaps in research ethics norms and practices | The technical progression of artificial intelligence (AI) research has been built on breakthroughs in fields such as computer science, statistics, and mathematics. However, in the past decade AI researchers have increasingly looked to the social sciences, turning to human interactions to solve the challenges of model development. Paying crowdsourcing workers to generate or curate data, or ‘data enrichment’, has become indispensable for many areas of AI research, from natural language processing to inverse reinforcement learning. Other fields that routinely interact with crowdsourcing workers, such as Psychology, have developed common governance requirements and norms to ensure research is undertaken ethically. This study explores how, and to what extent, comparable research ethics requirements and norms have developed for AI research and data enrichment. We focus on the approach taken by two leading AI conferences: ICLR and NeurIPS. In a longitudinal study of accepted papers, and a comparison with Springer journal articles and Psychology papers, this work finds that ICLR and NeurIPS have established protocols for human data collection which are inconsistently followed by authors. Whilst Psychology papers engaging with crowdsourcing workers frequently disclose ethics reviews, payment data, demographic data and other information, such disclosures are far less common in leading AI conferences despite similar guidance. The work concludes with hypotheses to explain these gaps in research ethics practices and considerations for its implications. | Reject |
ICLR.cc/2023/Conference | An Empirical Study on Anomaly detection Using Density Based and Representative Based Clustering algorithms | In data mining, and statistics, anomaly detection is the process of finding data patterns (outcomes, values, or observations) that deviate from the rest of the d other observations or outcomes. Anomaly detection is heavily used in solving real-world problems in many application domains like medicine, finance, cybersecurity, banking, networking, transportation, and military surveillance for enemy activities, but not limited to only those fields. In this paper, we present an empirical study of unsupervised anomaly detection techniques such as DBSCAN, DBSCAN++ (with uniform initialization, k-center initialization, uniform with approximate neighbor initialization, and k-center with approximate neighbor initialization), and k-means --(minus minus) algorithms on six benchmark imbalanced datasets. Findings from our in-depth empirical study show that k-means -- is a robust than DBSCAN, and DBSCAN++, in terms of the different evaluation measures (F1 score, False alarm rate, adjusted rand index, and Jaccard coefficient), and running time. We also observe that DBSCAN performs very well on datasets with fewer number of data points. | Withdrawn |
ICLR.cc/2019/Conference | Unsupervised Exploration with Deep Model-Based Reinforcement Learning | Reinforcement learning (RL) often requires large numbers of trials to solve a single specific task. This is in sharp contrast to human and animal learning: humans and animals can use past experience to acquire an understanding about the world, which they can then use to perform new tasks with minimal additional learning. In this work, we study how an unsupervised exploration phase can be used to build up such prior knowledge, which can then be utilized in a second phase to perform new tasks, either directly without any additional exploration, or through minimal fine-tuning. A critical question with this approach is: what kind of knowledge should be transferred from the unsupervised phase to the goal-directed phase? We argue that model-based RL offers an appealing solution. By transferring models, which are task-agnostic, we can perform new tasks without any additional learning at all. However, this relies on having a suitable exploration method during unsupervised training, and a model-based RL method that can effectively utilize modern high-capacity parametric function classes, such as deep neural networks. We show that both challenges can be addressed by representing model-uncertainty, which can both guide exploration in the unsupervised phase and ensure that the errors in the model are not exploited by the planner in the goal-directed phase. We illustrate, on simple simulated benchmark tasks, that our method can perform various goal-directed skills on the first attempt, and can improve further with fine-tuning, exceeding the performance of alternative exploration methods. | Reject |
ICLR.cc/2023/Conference | A Deep Conjugate Direction Method for Iteratively Solving Linear Systems | We present a novel deep learning approach to approximate the solution of large, sparse, symmetric, positive-definite linear systems of equations.These systems arise from many problems in applied science, e.g., in numerical methods for partial differential equations. Algorithms for approximating the solution to these systems are often the bottleneck in problems that require their solution, particularly for modern applications that require many millions of unknowns. Indeed, numerical linear algebra techniques have been investigated for many decades to alleviate this computational burden. Recently, data-driven techniques have also shown promise for these problems. Motivated by the conjugate gradients algorithm that iteratively selects search directions for minimizing the matrix norm of the approximation error, we design an approach that utilizes a deep neural network to accelerate convergence via data-driven improvement of the search directions.
Our method leverages a carefully chosen convolutional network to approximate the action of the inverse of the linear operator up to an arbitrary constant. We train the network using unsupervised learning with a loss function equal to the $L^2$ difference between an input and the system matrix times the network evaluation, where the unspecified constant in the approximate inverse is accounted for.
We demonstrate the efficacy of our approach on spatially discretized Poisson equations with millions of degrees of freedom arising in computational fluid dynamics applications.Unlike state-of-the-art learning approaches, our algorithm is capable of reducing the linear system residual to a given tolerance in a small number of iterations, independent of the problem size.Moreover, our method generalizes effectively to various systems beyond those encountered during training. | Reject |
ICLR.cc/2022/Conference | AlphaZero-based Proof Cost Network to Aid Game Solving | The AlphaZero algorithm learns and plays games without hand-crafted expert knowledge. However, since its objective is to play well, we hypothesize that a better objective can be defined for the related but separate task of solving games. This paper proposes a novel approach to solving problems by modifying the training target of the AlphaZero algorithm, such that it prioritizes solving the game quickly, rather than winning. We train a Proof Cost Network (PCN), where proof cost is a heuristic that estimates the amount of work required to solve problems. This matches the general concept of the so-called proof number from proof number search, which has been shown to be well-suited for game solving. We propose two specific training targets. The first finds the shortest path to a solution, while the second estimates the proof cost. We conduct experiments on solving 15x15 Gomoku and 9x9 Killall-Go problems with both MCTS-based and FDFPN solvers. Comparisons between using AlphaZero networks and PCN as heuristics show that PCN can solve more problems. | Accept (Poster) |
ICLR.cc/2023/Conference | Does Dataset Lottery Ticket Hypothesis Exist? | Tuning hyperparameters and exploring the suitable training schemes for the self-supervised models are usually expensive and resource-consuming, especially on large-scale datasets like ImageNet-1K. Critically, this means only a few establishments (e.g., Google, Meta, etc.) have the ability to afford the heavy experiments on this task, which seriously hinders more engagement and better development of this area. An ideal situation is that there exists a subset from the full large-scale dataset, the subset can correctly reflect the performance distinction when performing different training frameworks, hyper-parameters, etc. This new training manner will substantially decrease resource requirements and improve the computational performance of ablations without compromising accuracy on the full dataset. We formulate this interesting problem as the dataset lottery ticket hypothesis and the target subsets as the winning tickets.
In this work, we analyze this problem through finding out partial empirical data on the class dimension that has a consistent {\em Empirical Risk Trend} as the full observed dataset. We also examine multiple solutions, including (i) a uniform selection scheme that has been widely used in literature; (ii) subsets by involving prior knowledge, for instance, using the sorted per-class performance of the strong supervised model to identify the desired subset, WordNet Tree on hierarchical semantic classes, etc., for generating the target winning tickets.
We verify this hypothesis on the self-supervised learning task across a variety of recent mainstream methods, such as MAE, DINO, MoCo-V1/V2, etc., with different backbones like ResNet and Vision Transformers. The supervised classification task is also examined as an extension. We conduct extensive experiments for training more than 2K self-supervised models on the large-scale ImageNet-1K and its subsets by 1.5M GPU hours, to scrupulously deliver our discoveries and demonstrate our conclusions. According to our experimental results, the winning tickets (subsets) that we find behave consistently to the original dataset, which generally can benefit many experimental studies and ablations, saving 10x of training time and resources for the hyperparameter tuning and other ablation studies. | Reject |
ICLR.cc/2021/Conference | Average Reward Reinforcement Learning with Monotonic Policy Improvement | In continuing control tasks, an agent’s average reward per time step is a more natural performance measure compared to the commonly used discounting framework since it can better capture an agent’s long-term behavior. We derive a novel lower bound on the difference of the long-term average reward for two policies. The lower bound depends on the average divergence between the policies and on the so-called Kemeny constant, which measures to what degree the unichain Markov chains associated with the policies are well-connected. We also show that previous work based on the discounted return (Schulman et al., 2015; Achiam et al., 2017) results in a non-meaningful lower bound in the average reward setting. Based on our lower bound, we develop an iterative procedure which produces a sequence of monotonically improved policies for the average reward criterion. When combined with Deep Reinforcement Learning (DRL) methods, the procedure leads to scalable and efficient algorithms for maximizing the agent’s average reward performance.Empirically we demonstrate the effectiveness of our method on continuing control tasks and show how discounting can lead to unsatisfactory performance. | Reject |
ICLR.cc/2021/Conference | Efficient Empowerment Estimation for Unsupervised Stabilization | Intrinsically motivated artificial agents learn advantageous behavior without externally-provided rewards. Previously, it was shown that maximizing mutual information between agent actuators and future states, known as the empowerment principle, enables unsupervised stabilization of dynamical systems at upright positions, which is a prototypical intrinsically motivated behavior for upright standing and walking. This follows from the coincidence between the objective of stabilization and the objective of empowerment. Unfortunately, sample-based estimation of this kind of mutual information is challenging. Recently, various variational lower bounds (VLBs) on empowerment have been proposed as solutions; however, they are often biased, unstable in training, and have high sample complexity. In this work, we propose an alternative solution based on a trainable representation of a dynamical system as a Gaussian channel, which allows us to efficiently calculate an unbiased estimator of empowerment by convex optimization. We demonstrate our solution for sample-based unsupervised stabilization on different dynamical control systems and show the advantages of our method by comparing it to the existing VLB approaches. Specifically, we show that our method has a lower sample complexity, is more stable in training, possesses the essential properties of the empowerment function, and allows estimation of empowerment from images. Consequently, our method opens a path to wider and easier adoption of empowerment for various applications. | Accept (Poster) |
ICLR.cc/2022/Conference | Maximum n-times Coverage for Vaccine Design | We introduce the maximum $n$-times coverage problem that selects $k$ overlays to maximize the summed coverage of weighted elements, where each element must be covered at least $n$ times. We also define the min-cost $n$-times coverage problem where the objective is to select the minimum set of overlays such that the sum of the weights of elements that are covered at least $n$ times is at least $\tau$. Maximum $n$-times coverage is a generalization of the multi-set multi-cover problem, is NP-complete, and is not submodular. We introduce two new practical solutions for $n$-times coverage based on integer linear programming and sequential greedy optimization. We show that maximum $n$-times coverage is a natural way to frame peptide vaccine design, and find that it produces a pan-strain COVID-19 vaccine design that is superior to 29 other published designs in predicted population coverage and the expected number of peptides displayed by each individual's HLA molecules. | Accept (Poster) |
ICLR.cc/2020/Conference | Macro Action Ensemble Searching Methodology for Deep Reinforcement Learning | In this paper, we propose to improve the performance of deep reinforcement learn- ing (DRL) methods by searching for a feasible macro action ensemble to augment the action space of an agent. A macro action ensemble is composed of multiple macro actions, which are typically defined as sequences of primitive actions. A well-defined macro action ensemble enables a DRL agent to achieve higher performance than conventional DRL methods on a variety of tasks. However, macro actions generated by previous approaches are either not necessarily promising, or limited to specific forms. As a result, in this study, we investigate a search- ing method to learn the macro action ensemble from the environment of interest. The proposed method is inspired by the concepts of neural architecture search techniques, which are capable of developing network architectures for different tasks. These search techniques, such as NASNet or MetaQNN, have been proven to generate high-performance neural network architectures in large search spaces. In order to search in large macro action ensemble spaces, we propose to embrace Deep Q-Learning to search the macro action ensemble space for a good ensemble. Our approach iteratively discovers new ensembles of macro actions with better performance on the learning task. The proposed method is able to search finite macro action ensemble spaces directly, that the other contemporary methods have yet to achieve. Our experimental results show that the scores attained by the policy trained with the discovered macro action ensemble outperforms those without it. Moreover, the policies using our macro action ensemble are more efficient in exploration and able to converge faster. We further perform a comprehensive set of ablative analyses to validate the proposed methodology. | Withdrawn |
ICLR.cc/2023/Conference | Exploring interactions between modalities for deepfake detection | As face forgery techniques have become more mature, the proliferation of deepfakes may threat the human society security. Although existing deepfake detection methods achieve a good performance for in-dataset evaluation, it still remains to be improved in the generalization abiltiy, where the representation of the imperceptible artifacts plays a significant role. In this paper, we propose an Interactive Two-Stream Network (ITSNet) to explore the discriminant inconsistency representation from the perspective of cross-modality. Specially, the patch-wise Decomposable Discrete Cosine Transform (DDCT) is adopted to extract fine-grained high-frequency clues and information from different modalities are communitcated with each other via a designed interaction module. To perceive the temporal inconsistency, we first develop a Short-term Embedding Module (SEM) to refine subtle local inconsistency representation between adjacent frames, and then a Long-term Embedding Module (LEM) is designed to further refine the erratic temporal inconsistency representation from the long-range perspective. Extensive experimental results conducted on three public datasets show that ITSNet outperforms the state-of-the-art methods both in terms of in-dataset and cross-dataset evaluations. | Withdrawn |
ICLR.cc/2019/Conference | Explaining Neural Networks Semantically and Quantitatively | This paper presents a method to explain the knowledge encoded in a convolutional neural network (CNN) quantitatively and semantically. How to analyze the specific rationale of each prediction made by the CNN presents one of key issues of understanding neural networks, but it is also of significant practical values in certain applications. In this study, we propose to distill knowledge from the CNN into an explainable additive model, so that we can use the explainable model to provide a quantitative explanation for the CNN prediction. We analyze the typical bias-interpreting problem of the explainable model and develop prior losses to guide the learning of the explainable additive model. Experimental results have demonstrated the effectiveness of our method. | Withdrawn |
ICLR.cc/2022/Conference | Federated Distillation of Natural Language Understanding with Confident Sinkhorns | Enhancing the user experience is an essential task for application service providers. For instance, two users living wide apart may have different tastes of food. A food recommender mobile application installed on an edge device might want to learn from user feedback (reviews) to satisfy the client's needs pertaining to distinct domains. Retrieving user data comes at the cost of privacy while asking for model parameters trained on a user device becomes space inefficient at a large scale. In this work, we propose an approach to learn a central (global) model from the federation of (local) models which are trained on user-devices, without disclosing the local data or model parameters to the server. We propose a federation mechanism for the problems with natural similarity metric between the labels which commonly appear in natural language understanding (NLU) tasks. To learn the global model, the objective is to minimize the optimal transport cost of the global model's predictions from the confident sum of soft-targets assigned by local models. The confidence (a model weighting scheme) score of a model is defined as the L2 distance of a model's prediction from its probability bias. The method improves the global model's performance over the baseline designed on three NLU tasks with intrinsic label space semantics, i.e., fine-grained sentiment analysis, emotion recognition in conversation, and natural language inference. | Withdrawn |
ICLR.cc/2020/Conference | Continual Density Ratio Estimation (CDRE): A new method for evaluating generative models in continual learning | We propose a new method Continual Density Ratio Estimation (CDRE), which can estimate density ratios between a target distribution of real samples and a distribution of samples generated by a model while the model is changing over time and the data of the target distribution is not available after a certain time point. This method perfectly fits the setting of continual learning, in which one model is supposed to learn different tasks sequentially and the most crucial restriction is that model has none or very limited access to the data of all learned tasks. Through CDRE, we can evaluate generative models in continual learning using f-divergences. To the best of our knowledge, there is no existing method that can evaluate generative models under the setting of continual learning without storing real samples from the target distribution. | Reject |
ICLR.cc/2023/Conference | Training via Confidence Ranking | Model evolution and constant available data are two common phenomenon in large-scale real-world machine learning application, e.g. ads and recommendation system. To adapt, real-world system typically operates both retraining with all available data and \textit{online-learning} with recent available data to update models periodically with the goal of better serving performance for future data. However, if model and data evolution results in a vastly different training manner, it may induce negative impact on online A/B platform. In this paper, we propose a novel framework, named Confidence Ranking, to design optimization objective as a ranking function with two different models. Our confidence ranking loss allows directly optimizing the logits output for different convex surrogate function of metrics, e.g. AUC and Accuracy depending on the target tasks and datasets. Armed with our proposed methods, our experiments show that the confidence ranking loss can outperform for test-set performance on CTR prediction and model compression with various setting against the knowledge distillation baselines. | Withdrawn |
ICLR.cc/2018/Conference | A Classification-Based Perspective on GAN Distributions | A fundamental, and still largely unanswered, question in the context of Generative Adversarial Networks (GANs) is whether GANs are actually able to capture the key characteristics of the datasets they are trained on. The current approaches to examining this issue require significant human supervision, such as visual inspection of sampled images, and often offer only fairly limited scalability. In this paper, we propose new techniques that employ classification-based perspective to evaluate synthetic GAN distributions and their capability to accurately reflect the essential properties of the training data. These techniques require only minimal human supervision and can easily be scaled and adapted to evaluate a variety of state-of-the-art GANs on large, popular datasets. They also indicate that GANs have significant problems in reproducing the more distributional properties of the training dataset. In particular, the diversity of such synthetic data is orders of magnitude smaller than that of the original data. | Reject |
ICLR.cc/2023/Conference | MAT: Mixed-Strategy Game of Adversarial Training in Fine-tuning | Fine-tuning large-scale models from pre-trained checkpoints has been demonstrated effective for various natural language processing (NLP) tasks. Previous works reveal that leveraging adversarial training methods during the fine-tuning stage significantly enhances the generalization and robustness of the models. However, from the perspective of optimization, the previous adversarial training methods suffer from converging onto local optima due to the non-convexity of the objective. In this work, we reformulate the adversarial training in the view of mixed strategy in game theory and incorporate full strategy space to avoid trapping in local stationarity. Methodologically, we derive the Nash equilibrium of mixed-strategy for adversarial training using entropy mirror descent to establish a novel mixed-strategy adversarial training algorithm (MAT). Numerically, to verify the effectiveness of MAT, we conducted extensive benchmark experiments over the large-scale pre-trained models such as BERT and RoBERTa. The experimental results show that MAT outperforms the previous state-of-the-art on both GLUE and ANLI benchmarks in terms of generalization and robustness. | Withdrawn |
ICLR.cc/2022/Conference | Learning sparse DNNs with soft thresholding of weights during training | This paper proposes a new and simple way of training sparse neural networks. Our method is based on a differentiation of the forward and backward paths: the weights in the forward path are a thresholded version of the weights maintained in the backward path. This decoupling allows for micro-updates, produced by gradient descent, to stack up, leading to the possible re-activation of weights that were set to zero in earlier training steps. At the end of training, links with zero weights are pruned away.
Additional critical specificities of our approach lie (i) in the progressive increase of the zeroed weight ratio along the training, and (ii) in the use of soft-thresholding rather than hard-tresholding to derive the forward-path weights from the ones maintained in the backward path.
At constant accuracy, our approach reduces the number of training cycles to 1 compared to the state-of-the-art recursive pruning methods. At high pruning rates, it also improves the model accuracy compared to other single cycle pruning approaches (66.18% top-1 accuracy when training a ResNet-50 on ImageNet at 98% sparsity).
| Withdrawn |
ICLR.cc/2018/Conference | Prototype Matching Networks for Large-Scale Multi-label Genomic Sequence Classification | One of the fundamental tasks in understanding genomics is the problem of predicting Transcription Factor Binding Sites (TFBSs). With more than hundreds of Transcription Factors (TFs) as labels, genomic-sequence based TFBS prediction is a challenging multi-label classification task. There are two major biological mechanisms for TF binding: (1) sequence-specific binding patterns on genomes known as “motifs” and (2) interactions among TFs known as co-binding effects. In this paper, we propose a novel deep architecture, the Prototype Matching Network (PMN) to mimic the TF binding mechanisms. Our PMN model automatically extracts prototypes (“motif”-like features) for each TF through a novel prototype-matching loss. Borrowing ideas from few-shot matching models, we use the notion of support set of prototypes and an LSTM to learn how TFs interact and bind to genomic sequences. On a reference TFBS dataset with 2.1 million genomic sequences, PMN significantly outperforms baselines and validates our design choices empirically. To our knowledge, this is the first deep learning architecture that introduces prototype learning and considers TF-TF interactions for large scale TFBS prediction. Not only is the proposed architecture accurate, but it also models the underlying biology. | Reject |
ICLR.cc/2021/Conference | Faster Training of Word Embeddings | Word embeddings have gained increasing popularity in the recent years due to the Word2vec library and its extension fastText that uses subword information. In this paper, we aim at improving the execution speed of fastText training on homogeneous multi- and manycore CPUs while maintaining accuracy. We present a novel open-source implementation that flexibly incorporates various algorithmic variants including negative sample sharing, batched updates, and a byte-pair encoding-based alternative for subword units. We build these novel variants over a fastText implementation that we carefully optimized for the architecture, memory hierarchy, and parallelism of current manycore CPUs. Our experiments on three languages demonstrate 3-20x speed-up in training time at competitive semantic and syntactic accuracy. | Reject |
ICLR.cc/2018/Conference | A Neural Method for Goal-Oriented Dialog Systems to interact with Named Entities | Many goal-oriented dialog tasks, especially ones in which the dialog system has to interact with external knowledge sources such as databases, have to handle a large number of Named Entities (NEs). There are at least two challenges in handling NEs using neural methods in such settings: individual NEs may occur only rarely making it hard to learn good representations of them, and many of the Out Of Vocabulary words that occur during test time may be NEs. Thus, the need to interact well with these NEs has emerged as a serious challenge to building neural methods for goal-oriented dialog tasks. In this paper, we propose a new neural method for this problem, and present empirical evaluations on a structured Question answering task and three related goal-oriented dialog tasks that show that our proposed method can be effective in interacting with NEs in these settings. | Reject |
ICLR.cc/2023/Conference | Continuous pseudo-labeling from the start | Self-training (ST), or pseudo-labeling has sparked significant interest in the automatic speech recognition (ASR) community recently because of its success in harnessing unlabeled data. Unlike prior semi-supervised learning approaches that relied on iteratively regenerating pseudo-labels (PLs) from a trained model and using them to train a new model, recent state-of-the-art methods perform `continuous training' where PLs are generated using a very recent version of the model being trained. Nevertheless, these approaches still rely on bootstrapping the ST using an initial supervised learning phase where the model is trained on labeled data alone. We believe this has the potential for over-fitting to the labeled dataset in low resource settings and that ST from the start of training should reduce over-fitting. In this paper we show how we can do this by dynamically controlling the evolution of PLs during the training process in ASR. To the best of our knowledge, this is the first study that shows the feasibility of generating PLs from the very start of the training. We are able to achieve this using two techniques that avoid instabilities which lead to degenerate models that do not generalize. Firstly, we control the evolution of PLs through a curriculum that uses the online changes in PLs to control the membership of the cache of PLs and improve generalization. Secondly, we find that by sampling transcriptions from the predictive distribution, rather than only using the best transcription, we can stabilize training further. With these techniques, our ST models match prior works without an external language model. | Accept: poster |
ICLR.cc/2019/Conference | Fast adversarial training for semi-supervised learning | In semi-supervised learning, Bad GAN approach is one of the most attractive method due to the intuitional simplicity and powerful performances. Bad GAN learns a classifier with bad samples distributed on complement of the support of the input data. But Bad GAN needs additional architectures, a generator and a density estimation model, which involves huge computation and memory consumption cost. VAT is another good semi-supervised learning algorithm, which
utilizes unlabeled data to improve the invariance of the classifier with respect to perturbation of inputs. In this study, we propose a new method by combining the ideas of Bad GAN and VAT. The proposed method generates bad samples of high-quality by use of the adversarial training used in VAT. We give theoretical explanations why the adversarial training is good at both generating bad samples and semi-supervised learning. An advantage of the proposed method is to achieve the competitive performances with much fewer computations. We demonstrate advantages our method by various experiments with well known benchmark image datasets. | Reject |