Publications

How Transferable Are Features in Convolutional Neural Network Acoustic Models across Languages?
Jessica A.F. Thompson
Marc Schönwiesner
Yoshua Bengio
Daniel Willett
Characterization of the representations learned in intermediate layers of deep networks can provide valuable insight into the nature of a ta… (see more)sk and can guide the development of well-tailored learning strategies. Here we study convolutional neural network (CNN)-based acoustic models in the context of automatic speech recognition. Adapting a method proposed by [1], we measure the transferability of each layer between English, Dutch and German to assess their language-specificity. We observed three distinct regions of transferability: (1) the first two layers were entirely transferable between languages, (2) layers 2–8 were also highly transferable but we found some evidence of language specificity, (3) the subsequent fully connected layers were more language specific but could be successfully finetuned to the target language. To further probe the effect of weight freezing, we performed follow-up experiments using freeze-training [2]. Our results are consistent with the observation that CNNs converge ‘bottom up’ during training and demonstrate the benefit of freeze training, especially for transfer learning.
2019-04-30
IEEE International Conference on Acoustics, Speech, and Signal Processing (published)
doi.org
Compositional generalization in a deep seq2seq model by separating syntax and semantics
Jacob Russin
Jason Jo
R. O’Reilly
Yoshua Bengio
Standard methods in deep learning for natural language processing fail to capture the compositional structure of human language that allows … (see more)for systematic generalization outside of the training distribution. However, human learners readily generalize in this way, e.g. by applying known grammatical rules to novel words. Inspired by work in neuroscience suggesting separate brain systems for syntactic and semantic processing, we implement a modification to standard approaches in neural machine translation, imposing an analogous separation. The novel model, which we call Syntactic Attention, substantially outperforms standard methods in deep learning on the SCAN dataset, a compositional generalization task, without any hand-engineered features or additional supervision. Our work suggests that separating syntactic from semantic learning may be a useful heuristic for capturing compositional structure.
2019-04-21
ArXiv (preprint)
arxiv.org
arxiv.org
Continual Learning with Self-Organizing Maps
Pouya Bashivan
Martin Schrimpf
Robert Ajemian
Irina Rish
Matthew D Riemer
Yuhai Tu
Despite remarkable successes achieved by modern neural networks in a wide range of applications, these networks perform best in domain-speci… (see more)fic stationary environments where they are trained only once on large-scale controlled data repositories. When exposed to non-stationary learning environments, current neural networks tend to forget what they had previously learned, a phenomena known as catastrophic forgetting. Most previous approaches to this problem rely on memory replay buffers which store samples from previously learned tasks, and use them to regularize the learning on new ones. This approach suffers from the important disadvantage of not scaling well to real-life problems in which the memory requirements become enormous. We propose a memoryless method that combines standard supervised neural networks with self-organizing maps to solve the continual learning problem. The role of the self-organizing map is to adaptively cluster the inputs into appropriate task contexts - without explicit labels - and allocate network resources accordingly. Thus, it selectively routes the inputs in accord with previous experience, ensuring that past learning is maintained and does not interfere with current learning. Out method is intuitive, memoryless, and performs on par with current state-of-the-art approaches on standard benchmarks.
2019-04-18
ArXiv (preprint)
arxiv.org
arxiv.org
The PyTorch-Kaldi Speech Recognition Toolkit
Mirco Ravanelli
Titouan Parcollet
Yoshua Bengio
The availability of open-source software is playing a remarkable role in the popularization of speech recognition and deep learning. Kaldi, … (see more)for instance, is nowadays an established framework used to develop state-of-the-art speech recognizers. PyTorch is used to build neural networks with the Python language and has recently spawn tremendous interest within the machine learning community thanks to its simplicity and flexibility. The PyTorch-Kaldi project aims to bridge the gap between these popular toolkits, trying to inherit the efficiency of Kaldi and the flexibility of PyTorch. PyTorch-Kaldi is not only a simple interface between these software, but it embeds several useful features for developing modern speech recognizers. For instance, the code is specifically designed to naturally plug-in user-defined acoustic models. As an alternative, users can exploit several pre-implemented neural networks that can be customized using intuitive configuration files. PyTorch-Kaldi supports multiple feature and label streams as well as combinations of neural networks, enabling the use of complex neural architectures. The toolkit is publicly-released along with a rich documentation and is designed to properly work locally or on HPC clusters. Experiments, that are conducted on several datasets and tasks, show that PyTorch-Kaldi can effectively be used to develop modern state-of-the-art speech recognizers.
2019-04-16
ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (published)
doi.org
arxiv.org
GradMask: Reduce Overfitting by Regularizing Saliency
Becks Simpson
Francis Dutil
Yoshua Bengio
Joseph Paul Cohen
With too few samples or too many model parameters, overfitting can inhibit the ability to generalise predictions to new data. Within medical… (see more) imaging, this can occur when features are incorrectly assigned importance such as distinct hospital specific artifacts, leading to poor performance on a new dataset from a different institution without those features, which is undesirable. Most regularization methods do not explicitly penalize the incorrect association of these features to the target class and hence fail to address this issue. We propose a regularization method, GradMask, which penalizes saliency maps inferred from the classifier gradients when they are not consistent with the lesion segmentation. This prevents non-tumor related features to contribute to the classification of unhealthy samples. We demonstrate that this method can improve test accuracy between 1-3% compared to the baseline without GradMask, showing that it has an impact on reducing overfitting.
2019-04-14
MIDL.io/2019/Conference/Abstract (unknown)
openreview.net
openreview.net
Connecting Weighted Automata and Recurrent Neural Networks through Spectral Learning
Guillaume Rabusseau
Tianyu Li
Doina Precup
In this paper, we unravel a fundamental connection between weighted finite automata~(WFAs) and second-order recurrent neural networks~(2-RNN… (see more)s): in the case of sequences of discrete symbols, WFAs and 2-RNNs with linear activation functions are expressively equivalent. Motivated by this result, we build upon a recent extension of the spectral learning algorithm to vector-valued WFAs and propose the first provable learning algorithm for linear 2-RNNs defined over sequences of continuous input vectors. This algorithm relies on estimating low rank sub-blocks of the so-called Hankel tensor, from which the parameters of a linear 2-RNN can be provably recovered. The performances of the proposed method are assessed in a simulation study.
2019-04-10
Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics (published)
doi.org
proceedings.mlr.press
Distributional reinforcement learning with linear function approximation
Bellemare Marc-Emmanuel
Nicolas Roux
Pablo Samuel Castro
Subhodeep Moitra
Despite many algorithmic advances, our theoretical understanding of practical distributional reinforcement learning methods remains limited.… (see more) One exception is Rowland et al. (2018)'s analysis of the C51 algorithm in terms of the Cramer distance, but their results only apply to the tabular setting and ignore C51's use of a softmax to produce normalized distributions. In this paper we adapt the Cramer distance to deal with arbitrary vectors. From it we derive a new distributional algorithm which is fully Cramer-based and can be combined to linear function approximation, with formal guarantees in the context of policy evaluation. In allowing the model's prediction to be any real vector, we lose the probabilistic interpretation behind the method, but otherwise maintain the appealing properties of distributional approaches. To the best of our knowledge, ours is the first proof of convergence of a distributional algorithm combined with function approximation. Perhaps surprisingly, our results provide evidence that Cramer-based distributional methods may perform worse than directly approximating the value function.
2019-04-10
Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics (published)
proceedings.mlr.press
proceedings.mlr.press
Multitask Metric Learning: Theory and Algorithm
Boyu Wang
Hejia Zhang
Peng Liu
Zebang Shen
Joelle Pineau
In this paper, we study the problem of multitask metric learning (mtML). We first examine the generalization bound of the regularized mtML f… (see more)ormulation based on the notion of algorithmic stability, proving the convergence rate of mtML and revealing the trade-off between the tasks. Moreover, we also establish the theoretical connection between the mtML, single-task learning and pooling-task learning approaches. In addition, we present a novel boosting-based mtML (mt-BML) algorithm, which scales well with the feature dimension of the data. Finally, we also devise an efficient second-order Riemannian retraction operator which is tailored specifically to our mt-BML algorithm. It produces a low-rank solution of mtML to reduce the model complexity, and may also improve generalization performances. Extensive evaluations on several benchmark data sets verify the effectiveness of our learning algorithm.
2019-04-10
International Conference on Artificial Intelligence and Statistics (published)
proceedings.mlr.press
proceedings.mlr.press
Negative Momentum for Improved Game Dynamics
Gauthier Gidel
Reyhane Askari Hemmat
Mohammad Pezeshki
Gabriel Huang
Rémi Le Priol
Simon Lacoste-Julien
Ioannis Mitliagkas
Games generalize the single-objective optimization paradigm by introducing different objective functions for different players. Differentiab… (see more)le games often proceed by simultaneous or alternating gradient updates. In machine learning, games are gaining new importance through formulations like generative adversarial networks (GANs) and actor-critic systems. However, compared to single-objective optimization, game dynamics are more complex and less understood. In this paper, we analyze gradient-based methods with momentum on simple games. We prove that alternating updates are more stable than simultaneous updates. Next, we show both theoretically and empirically that alternating gradient updates with a negative momentum term achieves convergence in a difficult toy adversarial problem, but also on the notoriously difficult to train saturating GANs.
2019-04-10
Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics (published)
doi.org
proceedings.mlr.press
Reinforced Imitation in Heterogeneous Action Space
Konrad Żołna
Negar Rostamzadeh
Yoshua Bengio
Sungjin Ahn
Pedro O. Pinheiro
Imitation learning is an effective alternative approach to learn a policy when the reward function is sparse. In this paper, we consider a c… (see more)hallenging setting where an agent and an expert use different actions from each other. We assume that the agent has access to a sparse reward function and state-only expert observations. We propose a method which gradually balances between the imitation learning cost and the reinforcement learning objective. In addition, this method adapts the agent's policy based on either mimicking expert behavior or maximizing sparse reward. We show, through navigation scenarios, that (i) an agent is able to efficiently leverage sparse rewards to outperform standard state-only imitation learning, (ii) it can learn a policy even when its actions are different from the expert, and (iii) the performance of the agent is not bounded by that of the expert, due to the optimized usage of sparse rewards.
2019-04-05
ArXiv (preprint)
arxiv.org
arxiv.org
A Survey on Practical Applications of Multi-Armed and Contextual Bandits
Djallel Bouneffouf
Irina Rish
In recent years, multi-armed bandit (MAB) framework has attracted a lot of attention in various applications, from recommender systems and i… (see more)nformation retrieval to healthcare and finance, due to its stellar performance combined with certain attractive properties, such as learning from less feedback. The multi-armed bandit field is currently flourishing, as novel problem settings and algorithms motivated by various practical applications are being introduced, building on top of the classical bandit problem. This article aims to provide a comprehensive review of top recent developments in multiple real-life applications of the multi-armed bandit. Specifically, we introduce a taxonomy of common MAB-based applications and summarize state-of-art for each of those domains. Furthermore, we identify important current trends and provide new perspectives pertaining to the future of this exciting and fast-growing field.
2019-04-01
ArXiv (preprint)
arxiv.org
arxiv.org
Gated Orthogonal Recurrent Units: On Learning to Forget
Li Jing
Caglar Gulçehre
John Peurifoy
Yichen Shen
Max Tegmark
Marin Soljacic
Yoshua Bengio
We present a novel recurrent neural network (RNN)–based model that combines the remembering ability of unitary evolution RNNs with the abi… (see more)lity of gated RNNs to effectively forget redundant or irrelevant information in its memory. We achieve this by extending restricted orthogonal evolution RNNs with a gating mechanism similar to gated recurrent unit RNNs with a reset gate and an update gate. Our model is able to outperform long short-term memory, gated recurrent units, and vanilla unitary or orthogonal RNNs on several long-term-dependency benchmark tasks. We empirically show that both orthogonal and unitary RNNs lack the ability to forget. This ability plays an important role in RNNs. We provide competitive results along with an analysis of our model on many natural sequential tasks, including question answering, speech spectrum prediction, character-level language modeling, and synthetic tasks that involve long-term dependencies such as algorithmic, denoising, and copying tasks.
2019-03-31
Neural Computation (published)
doi.org
arxiv.org