Paper Group NANR 126
Guided Adaptive Credit Assignment for Sample Efficient Policy Optimization. Reasoning-Aware Graph Convolutional Network for Visual Question Answering. Attention on Abstract Visual Reasoning. CLEVRER: Collision Events for Video Representation and Reasoning. Deep End-to-end Unsupervised Anomaly Detection. Attentive Weights Generation for Few Shot Lea …
Guided Adaptive Credit Assignment for Sample Efficient Policy Optimization
| Title | Guided Adaptive Credit Assignment for Sample Efficient Policy Optimization |
| Authors | Anonymous |
| Abstract | Policy gradient methods have achieved remarkable successes in solving challenging reinforcement learning problems. However, it still often suffers from sparse reward tasks, which leads to poor sample efficiency during training. In this work, we propose a guided adaptive credit assignment method to do effectively credit assignment for policy gradient methods. Motivated by entropy regularized policy optimization, our method extends the previous credit assignment methods by introducing more general guided adaptive credit assignment(GACA). The benefit of GACA is a principled way of utilizing off-policy samples. The effectiveness of proposed algorithm is demonstrated on the challenging \textsc{WikiTableQuestions} and \textsc{WikiSQL} benchmarks and an instruction following environment. The task is generating action sequences or program sequences from natural language questions or instructions, where only final binary success-failure execution feedback is available. Empirical studies show that our method significantly improves the sample efficiency of the state-of-the-art policy optimization approaches. |
| Tasks | Policy Gradient Methods |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SyxBgkBFPS |
| https://openreview.net/pdf?id=SyxBgkBFPS | |
| PWC | https://paperswithcode.com/paper/guided-adaptive-credit-assignment-for-sample |
| Repo | |
| Framework | |
Reasoning-Aware Graph Convolutional Network for Visual Question Answering
| Title | Reasoning-Aware Graph Convolutional Network for Visual Question Answering |
| Authors | Anonymous |
| Abstract | Relational reasoning methods based on graph networks are currently state-of-the-art models for Visual Question Answering (VQA) tasks involving real images. Although graph networks are used in these models to enrich visual representations by encoding question-adaptive inter-object relations, these simple graph networks is arguably insufficient to perform visual reasoning for VQA tasks. In this paper, we propose a Reasoning-Aware Graph Convolutional Networks (RA-GCN) that goes one step further towards visual reasoning for GCNs. Our first contribution is the introduction of visual reasoning ability into conventional GCNs. Secondly, we strengthen the expressive power of GCNs via introducing node-sensitive kernel parameters based on edge features to address the limitation of shared transformation matrix for each node in GCNs. Finally, we provide a novel iterative reasoning network architecture for solving VQA task via embedding the RA-GCN module into an iterative process. We evaluate our model on the VQA-CP v2, GQA and Clevr dataset. Our final RA-GCN network successfully achieves state-of-the-art accuracy which is 42.3% on the VQA-CP v2, and highly competitive 62.4% accuracy on the GQA, as well as 90.0% on val split of Clevr dataset. |
| Tasks | Question Answering, Relational Reasoning, Visual Question Answering, Visual Reasoning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SkgPvlSYwS |
| https://openreview.net/pdf?id=SkgPvlSYwS | |
| PWC | https://paperswithcode.com/paper/reasoning-aware-graph-convolutional-network |
| Repo | |
| Framework | |
Attention on Abstract Visual Reasoning
| Title | Attention on Abstract Visual Reasoning |
| Authors | Anonymous |
| Abstract | Attention mechanisms have been boosting the performance of deep learning models on a wide range of applications, ranging from speech understanding to program induction. However, despite experiments from psychology which suggest that attention plays an essential role in visual reasoning, the full potential of attention mechanisms has so far not been explored to solve abstract cognitive tasks on image data. In this work, we propose a hybrid network architecture, grounded on self-attention and relational reasoning. We call this new model Attention Relation Network (ARNe). ARNe combines features from the recently introduced Transformer and the Wild Relation Network (WReN). We test ARNe on the Procedurally Generated Matrices (PGMs) datasets for abstract visual reasoning. ARNe excels the WReN model on this task by 11.28 ppt. Relational concepts between objects are efficiently learned demanding only 35% of the training samples to surpass reported accuracy of the base line model. Our proposed hybrid model, represents an alternative on learning abstract relations using self-attention and demonstrates that the Transformer network is also well suited for abstract visual reasoning. |
| Tasks | Relational Reasoning, Visual Reasoning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Bkel1krKPS |
| https://openreview.net/pdf?id=Bkel1krKPS | |
| PWC | https://paperswithcode.com/paper/attention-on-abstract-visual-reasoning-1 |
| Repo | |
| Framework | |
CLEVRER: Collision Events for Video Representation and Reasoning
| Title | CLEVRER: Collision Events for Video Representation and Reasoning |
| Authors | Anonymous |
| Abstract | The ability to reason about temporal and causal events from videos lies at the core of human intelligence. Most video reasoning benchmarks, however, focus on pattern recognition from complex visual and language input, instead of on causal structure. We study the complementary problem, exploring the temporal and causal structures behind videos of objects with simple visual appearance. To this end, we introduce the CoLlision Events for Video REpresentation and Reasoning (CLEVRER) dataset, a diagnostic video dataset for systematic evaluation of computational models on a wide range of reasoning tasks. Motivated by the theory of human casual judgment, CLEVRER includes four types of question: descriptive (e.g., ‘what color’), explanatory (‘what’s responsible for’), predictive (‘what will happen next’), and counterfactual (‘what if’). We evaluate various state-of-the-art models for visual reasoning on our benchmark. While these models thrive on the perception-based task (descriptive), they perform poorly on the causal tasks (explanatory, predictive and counterfactual), suggesting that a principled approach for causal reasoning should incorporate the capability of both perceiving complex visual and language inputs, and understanding the underlying dynamics and causal relations. We also study an oracle model that explicitly combines these components via symbolic representations. CLEVRER will be made publicly available. |
| Tasks | Visual Reasoning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HkxYzANYDB |
| https://openreview.net/pdf?id=HkxYzANYDB | |
| PWC | https://paperswithcode.com/paper/clevrer-collision-events-for-video-1 |
| Repo | |
| Framework | |
Deep End-to-end Unsupervised Anomaly Detection
| Title | Deep End-to-end Unsupervised Anomaly Detection |
| Authors | Anonymous |
| Abstract | This paper proposes a novel method to detect anomalies in large datasets under a fully unsupervised setting. The key idea behind our algorithm is to learn the representation underlying normal data. To this end, we leverage the latest clustering technique suitable for handling high dimensional data. This hypothesis provides a reliable starting point for normal data selection. We train an autoencoder from the normal data subset, and iterate between hypothesizing normal candidate subset based on clustering and representation learning. The reconstruction error from the learned autoencoder serves as a scoring function to assess the normality of the data. Experimental results on several public benchmark datasets show that the proposed method outperforms state-of-the-art unsupervised techniques and is comparable to semi-supervised techniques in most cases. |
| Tasks | Anomaly Detection, Representation Learning, Unsupervised Anomaly Detection |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Bye3P1BYwr |
| https://openreview.net/pdf?id=Bye3P1BYwr | |
| PWC | https://paperswithcode.com/paper/deep-end-to-end-unsupervised-anomaly |
| Repo | |
| Framework | |
Attentive Weights Generation for Few Shot Learning via Information Maximization
| Title | Attentive Weights Generation for Few Shot Learning via Information Maximization |
| Authors | Anonymous |
| Abstract | Few shot image classification aims at learning a classifier from limited labeled data. Generating the classification weights has been applied in many meta-learning approaches for few shot image classification due to its simplicity and effectiveness. However, we argue that it is difficult to generate the exact and universal classification weights for all the diverse query samples from very few training samples. In this work, we introduce Attentive Weights Generation for few shot learning via Information Maximization (AWGIM), which addresses current issues by two novel contributions. i) AWGIM generates different classification weights for different query samples by letting each of query samples attends to the whole support set. ii) To guarantee the generated weights adaptive to different query sample, we re-formulate the problem to maximize the lower bound of mutual information between generated weights and query as well as support data. As far as we can see, this is the first attempt to unify information maximization into few shot learning. Both two contributions are proved to be effective in the extensive experiments and we show that AWGIM is able to achieve state-of-the-art performance on benchmark datasets. |
| Tasks | Few-Shot Image Classification, Few-Shot Learning, Image Classification, Meta-Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BJxpIJHKwB |
| https://openreview.net/pdf?id=BJxpIJHKwB | |
| PWC | https://paperswithcode.com/paper/attentive-weights-generation-for-few-shot |
| Repo | |
| Framework | |
Deep Auto-Deferring Policy for Combinatorial Optimization
| Title | Deep Auto-Deferring Policy for Combinatorial Optimization |
| Authors | Anonymous |
| Abstract | Designing efficient algorithms for combinatorial optimization appears ubiquitously in various scientific fields. Recently, deep reinforcement learning (DRL) frameworks have gained considerable attention as a new approach: they can automatically learn the design of a good solver without using any sophisticated knowledge or hand-crafted heuristic specialized for the target problem. However, the number of stages (until reaching the final solution) required by existing DRL solvers is proportional to the size of the input graph, which hurts their scalability to large-scale instances. In this paper, we seek to resolve this issue by proposing a novel design of DRL’s policy, coined auto-deferring policy (ADP), automatically stretching or shrinking its decision process. Specifically, it decides whether to finalize the value of each vertex at the current stage or defer to determine it at later stages. We apply the proposed ADP framework to the maximum independent set (MIS) problem, a prototype of NP-complete problems, under various scenarios. Our experimental results demonstrate significant improvement of ADP over the current state-of-the-art DRL scheme in terms of computational efficiency and approximation quality. The reported performance of our generic DRL scheme is also comparable with that of the state-of-the-art solvers specialized for MIS, e.g., ADP outperforms them for some graphs with millions of vertices. |
| Tasks | Combinatorial Optimization |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Hkexw1BtDr |
| https://openreview.net/pdf?id=Hkexw1BtDr | |
| PWC | https://paperswithcode.com/paper/deep-auto-deferring-policy-for-combinatorial |
| Repo | |
| Framework | |
Neural-Guided Symbolic Regression with Asymptotic Constraints
| Title | Neural-Guided Symbolic Regression with Asymptotic Constraints |
| Authors | Anonymous |
| Abstract | Symbolic regression is a type of discrete optimization problem that involves searching expressions that fit given data points. In many cases, other mathematical constraints about the unknown expression not only provide more information beyond just values at some inputs, but also effectively constrain the search space. We identify the asymptotic constraints of leading polynomial powers as the function approaches 0 and infinity as useful constraints and create a system to use them for symbolic regression. The first part of the system is a conditional expression generating neural network which preferentially generates expressions with the desired leading powers, producing novel expressions outside the training domain. The second part, which we call Neural-Guided Monte Carlo Tree Search, uses the network during a search to find an expression that conforms to a set of data points and desired leading powers. Lastly, we provide an extensive experimental validation on thousands of target expressions showing the efficacy of our system compared to exiting methods for finding unknown functions outside of the training set. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=S1gTAp4FDB |
| https://openreview.net/pdf?id=S1gTAp4FDB | |
| PWC | https://paperswithcode.com/paper/neural-guided-symbolic-regression-with-1 |
| Repo | |
| Framework | |
Learning to Learn Kernels with Variational Random Features
| Title | Learning to Learn Kernels with Variational Random Features |
| Authors | Anonymous |
| Abstract | Meta-learning for few-shot learning involves a meta-learner that acquires shared knowledge from a set of prior tasks to improve the performance of a base-learner on new tasks with a small amount of data. Kernels are commonly used in machine learning due to their strong nonlinear learning capacity, which have not yet been fully investigated in the meta-learning scenario for few-shot learning. In this work, we explore kernel approximation with random Fourier features in the meta-learning framework for few-shot learning. We propose learning adaptive kernels by meta variational random features (MetaVRF), which is formulated as a variational inference problem. To explore shared knowledge across diverse tasks, our MetaVRF deploys an LSTM inference network to generate informative features, which can establish kernels of highly representational power with low spectral sampling rates, while also being able to quickly adapt to specific tasks for improved performance. We evaluate MetaVRF on a variety of few-shot learning tasks for both regression and classification. Experimental results demonstrate that our MetaVRF can deliver much better or competitive performance than recent meta-learning algorithms. |
| Tasks | Few-Shot Learning, Meta-Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJebgkSFDB |
| https://openreview.net/pdf?id=rJebgkSFDB | |
| PWC | https://paperswithcode.com/paper/learning-to-learn-kernels-with-variational |
| Repo | |
| Framework | |
Using Logical Specifications of Objectives in Multi-Objective Reinforcement Learning
| Title | Using Logical Specifications of Objectives in Multi-Objective Reinforcement Learning |
| Authors | Anonymous |
| Abstract | In the multi-objective reinforcement learning (MORL) paradigm, the relative importance of each environment objective is often unknown prior to training, so agents must learn to specialize their behavior to optimize different combinations of environment objectives that are specified post-training. These are typically linear combinations, so the agent is effectively parameterized by a weight vector that describes how to balance competing environment objectives. However, many real world behaviors require non-linear combinations of objectives. Additionally, the conversion between desired behavior and weightings is often unclear. In this work, we explore the use of a language based on propositional logic with quantitative semantics–in place of weight vectors–for specifying non-linear behaviors in an interpretable way. We use a recurrent encoder to encode logical combinations of objectives, and train a MORL agent to generalize over these encodings. We test our agent in several grid worlds with various objectives and show that our agent can generalize to many never-before-seen specifications with performance comparable to single policy baseline agents. We also demonstrate our agent’s ability to generate meaningful policies when presented with novel specifications and quickly specialize to novel specifications. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJeuMREKwS |
| https://openreview.net/pdf?id=rJeuMREKwS | |
| PWC | https://paperswithcode.com/paper/using-logical-specifications-of-objectives-in-1 |
| Repo | |
| Framework | |
Progressive Compressed Records: Taking a Byte Out of Deep Learning Data
| Title | Progressive Compressed Records: Taking a Byte Out of Deep Learning Data |
| Authors | Anonymous |
| Abstract | Deep learning training accesses vast amounts of data at high velocity, posing challenges for datasets retrieved over commodity networks and storage devices. We introduce a way to dynamically reduce the overhead of fetching and transporting training data with a method we term Progressive Compressed Records (PCRs). PCRs deviate from previous formats by leveraging progressive compression to split each training example into multiple examples of increasingly higher fidelity, without adding to the total data size. Training examples of similar fidelity are grouped together, which reduces both the system overhead and data bandwidth needed to train a model. We show that models can be trained on aggressively compressed representations of the training data and still retain high accuracy, and that PCRs can enable a 2x speedup on average over baseline formats using JPEG compression. Our results hold across deep learning architectures for a wide range of datasets: ImageNet, HAM10000, Stanford Cars, and CelebA-HQ. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=S1e0ZlHYDB |
| https://openreview.net/pdf?id=S1e0ZlHYDB | |
| PWC | https://paperswithcode.com/paper/progressive-compressed-records-taking-a-byte |
| Repo | |
| Framework | |
ASGen: Answer-containing Sentence Generation to Pre-Train Question Generator for Scale-up Data in Question Answering
| Title | ASGen: Answer-containing Sentence Generation to Pre-Train Question Generator for Scale-up Data in Question Answering |
| Authors | Anonymous |
| Abstract | Numerous machine reading comprehension (MRC) datasets often involve manual annotation, requiring enormous human effort, and hence the size of the dataset remains significantly smaller than the size of the data available for unsupervised learning. Recently, researchers proposed a model for generating synthetic question-and-answer data from large corpora such as Wikipedia. This model is utilized to generate synthetic data for training an MRC model before fine-tuning it using the original MRC dataset. This technique shows better performance than other general pre-training techniques such as language modeling, because the characteristics of the generated data are similar to those of the downstream MRC data. However, it is difficult to have high-quality synthetic data comparable to human-annotated MRC datasets. To address this issue, we propose Answer-containing Sentence Generation (ASGen), a novel pre-training method for generating synthetic data involving two advanced techniques, (1) dynamically determining K answers and (2) pre-training the question generator on the answer-containing sentence generation task. We evaluate the question generation capability of our method by comparing the BLEU score with existing methods and test our method by fine-tuning the MRC model on the downstream MRC data after training on synthetic data. Experimental results show that our approach outperforms existing generation methods and increases the performance of the state-of-the-art MRC models across a range of MRC datasets such as SQuAD-v1.1, SQuAD-v2.0, KorQuAD and QUASAR-T without any architectural modifications to the original MRC model. |
| Tasks | Language Modelling, Machine Reading Comprehension, Question Answering, Question Generation, Reading Comprehension |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=H1lFsREYPS |
| https://openreview.net/pdf?id=H1lFsREYPS | |
| PWC | https://paperswithcode.com/paper/asgen-answer-containing-sentence-generation |
| Repo | |
| Framework | |
Continual Learning with Gated Incremental Memories for Sequential Data Processing
| Title | Continual Learning with Gated Incremental Memories for Sequential Data Processing |
| Authors | Anonymous |
| Abstract | The ability to learn over changing task distributions without forgetting previous knowledge, also known as continual learning, is a key enabler for scalable and trustworthy deployments of adaptive solutions. While the importance of continual learning is largely acknowledged in machine vision and reinforcement learning problems, this is mostly under-documented for sequence processing tasks. This work focuses on characterizing and quantitatively assessing the impact of catastrophic forgetting and task interference when dealing with sequential data in recurrent neural networks. We also introduce a general architecture, named Gated Incremental Memory, for augmenting recurrent models with continual learning skills, whose effectiveness is demonstrated through the benchmarks introduced in this paper. |
| Tasks | Continual Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HklliySFDS |
| https://openreview.net/pdf?id=HklliySFDS | |
| PWC | https://paperswithcode.com/paper/continual-learning-with-gated-incremental |
| Repo | |
| Framework | |
Revisiting Gradient Episodic Memory for Continual Learning
| Title | Revisiting Gradient Episodic Memory for Continual Learning |
| Authors | Anonymous |
| Abstract | Gradient Episodic Memory (GEM) is an effective model for continual learning, where each gradient update for the current task is formulated as a quadratic program problem with inequality constraints that alleviate catastrophic forgetting of previous tasks. However, practical use of GEM is impeded by several limitations: (1) the data examples stored in the episodic memory may not be representative of past tasks; (2) the inequality constraints appear to be rather restrictive for competing or conflicting tasks; (3) the inequality constraints can only avoid catastrophic forgetting but can not assure positive backward transfer. To address these issues, in this paper we aim at improving the original GEM model via three handy techniques without extra computational cost. Experiments on MNIST Permutations and incremental CIFAR100 datasets demonstrate that our techniques enhance the performance of GEM remarkably. On CIFAR100 the average accuracy is improved from 66.48% to 68.76%, along with the backward (knowledge) transfer growing from 1.38% to 4.03%. |
| Tasks | Continual Learning, Transfer Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=H1g79ySYvB |
| https://openreview.net/pdf?id=H1g79ySYvB | |
| PWC | https://paperswithcode.com/paper/revisiting-gradient-episodic-memory-for |
| Repo | |
| Framework | |
Robust Cross-lingual Embeddings from Parallel Sentences
| Title | Robust Cross-lingual Embeddings from Parallel Sentences |
| Authors | Anonymous |
| Abstract | Recent advances in cross-lingual word embeddings have primarily relied on mapping-based methods, which project pretrained word embeddings from different languages into a shared space through a linear transformation. However, these approaches assume word embedding spaces are isomorphic between different languages, which has been shown not to hold in practice (Søgaard et al., 2018), and fundamentally limits their performance. This motivates investigating joint learning methods which can overcome this impediment, by simultaneously learning embeddings across languages via a cross-lingual term in the training objective. Given the abundance of parallel data available (Tiedemann, 2012), we propose a bilingual extension of the CBOW method which leverages sentence-aligned corpora to obtain robust cross-lingual word and sentence representations. Our approach significantly improves cross-lingual sentence retrieval performance over all other approaches, as well as convincingly outscores mapping methods while maintaining parity with jointly trained methods on word-translation. It also achieves parity with a deep RNN method on a zero-shot cross-lingual document classification task, requiring far fewer computational resources for training and inference. As an additional advantage, our bilingual method also improves the quality of monolingual word vectors despite training on much smaller datasets. We make our code and models publicly available. |
| Tasks | Cross-Lingual Document Classification, Document Classification, Word Embeddings |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SJlJSaEFwS |
| https://openreview.net/pdf?id=SJlJSaEFwS | |
| PWC | https://paperswithcode.com/paper/robust-cross-lingual-embeddings-from-parallel |
| Repo | |
| Framework | |
