Publications

Continual Learning in Deep Networks: an Analysis of the Last Layer
Timothee LESORT
Thomas George
Irina Rish
We study how different output layers in a deep neural network learn and forget in continual learning settings. The following three factors… (voir plus) can affect catastrophic forgetting in the output layer: (1) weights modifications, (2) interference, and (3) projection drift. In this paper, our goal is to provide more insights into how changing the output layers may address (1) and (2). Some potential solutions to those issues are proposed and evaluated here in several continual learning scenarios. We show that the best-performing type of the output layer depends on the data distribution drifts and/or the amount of data available. In particular, in some cases where a standard linear layer would fail, it turns out that changing parameterization is sufficient in order to achieve a significantly better performance, whithout introducing a continual-learning algorithm and instead using the standard SGD to train a model. Our analysis and results shed light on the dynamics of the output layer in continual learning scenarios, and suggest a way of selecting the best type of output layer for a given scenario.
2021-06-02
ArXiv (prépublication)
openreview.net
openreview.net
Enquire One’s Parent and Child Before Decision: Fully Exploit Hierarchical Structure for Self-Supervised Taxonomy Expansion
Suyuchen Wang
Ruihui Zhao
X. T. Chen
Yefeng Zheng
Bang Liu
Taxonomy is a hierarchically structured knowledge graph that plays a crucial role in machine intelligence. The taxonomy expansion task aims … (voir plus)to find a position for a new term in an existing taxonomy to capture the emerging knowledge in the world and keep the taxonomy dynamically updated. Previous taxonomy expansion solutions neglect valuable information brought by the hierarchical structure and evaluate the correctness of merely an added edge, which downgrade the problem to node-pair scoring or mini-path classification. In this paper, we propose the Hierarchy Expansion Framework (HEF), which fully exploits the hierarchical structure’s properties to maximize the coherence of expanded taxonomy. HEF makes use of taxonomy’s hierarchical structure in multiple aspects: i) HEF utilizes subtrees containing most relevant nodes as self-supervision data for a complete comparison of parental and sibling relations; ii) HEF adopts a coherence modeling module to evaluate the coherence of a taxonomy’s subtree by integrating hypernymy relation detection and several tree-exclusive features; iii) HEF introduces the Fitting Score for position selection, which explicitly evaluates both path and level selections and takes full advantage of parental relations to interchange information for disambiguation and self-correction. Extensive experiments show that by better exploiting the hierarchical structure and optimizing taxonomy’s coherence, HEF vastly surpasses the prior state-of-the-art on three benchmark datasets by an average improvement of 46.7% in accuracy and 32.3% in mean reciprocal rank.
2021-06-02
Proceedings of the Web Conference 2021 (publié)
doi.org
arxiv.org
Brainhack: Developing a culture of open, inclusive, community-driven neuroscience
Rémi Gau
Stephanie Noble
Katja Heuer
Katherine L. Bottenhorn
Isil Poyraz Bilgin
Yufang Yang
Julia M. Huntenburg
Johanna Bayer
Richard A. I. Bethlehem
Shawn A Rhoads
Christoph Vogelbacher
Valentina Borghesani
Elizabeth Levitis
Hao-Ting Wang
Sofie Van Den Bossche
Xenia Kobeleva
Jon Haitz Legarreta
Samuel Guay
Melvin Selim Atay
Gael P. Varoquaux … (voir 80 de plus)
Dorien Huijser
Malin Sandström
Peer Herholz
Samuel A. Nastase
AmanPreet Badhwar
Guillaume Dumas
Simon Schwab
Stefano Moia
Michael Dayan
Yasmine Bassil
Paula P. Brooks
Matteo Mancini
James M. Shine
David O’Connor
Xihe Xie
Davide Poggiali
Patrick Friedrich
Anibal Sólon Heinsfeld
Lydia Riedl
Roberto Toro
César Caballero‐Gaudes
Anders Eklund
Kelly Garner
Christopher Nolan
Damion V. Demeter
Fernando A. Barrios
Junaid S. Merchant
Elizabeth A. McDevitt
Robert Oostenveld
R. Cameron Craddock
Ariel Rokem
Andrew Doyle
Satrajit Ghosh
Aki Nikolaidis
Olivia W. Stanley
Eneko Uruñuela
Nasim Anousheh
Aurina Arnatkevičiūtė
Guillaume Auzias
Dipankar Bachar
Élise Bannier
Ruggero Basanisi
Arshitha Basavaraj
Marco Bedini
Pierre Bellec
R. Austin Benn
Kathryn Berluti
Steffen Bollmann
Saskia Bollmann
Claire Bradley
Jesse A. Brown
Augusto Buchweitz
Patrick Callahan
Micaela Y. Chan
Bramsh Q. Chandio
Theresa W Cheng
Sidhant Chopra
Ai Wern Chung
Thomas Close
Etienne Combrisson
Giorgia Cona
R. Todd Constable
Claire Cury
Kamalaker Dadi
Pablo F. Damasceno
Samir Das
Fabrizio De Vico Fallani
Krista DeStasio
Erin W. Dickie
Lena Dorfschmidt
Eugene Duff
Elizabeth DuPré
Sarah L. Dziura
Nathália Bianchini Esper
Oscar Estéban
Shreyas Fadnavis
Guillaume Flandin
Jessica Flannery
John C. Flournoy
Stephanie J. Forkel

Brainhack is an innovative meeting format that promotes scientific collaboration and education in an open and inclusive environment. Depa… (voir plus)rting from the formats of typical scientific workshops, these events are based on grassroots projects and training, and foster open and reproducible scientific practices. We describe here the multifaceted, lasting benefits of Brainhacks for individual participants, particularly early career researchers. We further highlight the unique contributions that Brainhacks can make to the research community, contributing to scientific progress by complementing opportunities available in conventional formats.

2021-05-31
Neuron (publié)
doi.org
ER-AE: Differentially Private Text Generation for Authorship Anonymization
Haohan Bo
Steven H. H. Ding
Benjamin C. M. Fung
Farkhund Iqbal
Most of privacy protection studies for textual data focus on removing explicit sensitive identifiers. However, personal writing style, as a … (voir plus)strong indicator of the authorship, is often neglected. Recent studies, such as SynTF, have shown promising results on privacy-preserving text mining. However, their anonymization algorithm can only output numeric term vectors which are difficult for the recipients to interpret. We propose a novel text generation model with a two-set exponential mechanism for authorship anonymization. By augmenting the semantic information through a REINFORCE training reward function, the model can generate differentially private text that has a close semantic and similar grammatical structure to the original text while removing personal traits of the writing style. It does not assume any conditioned labels or paralleled text data for training. We evaluate the performance of the proposed model on the real-life peer reviews dataset and the Yelp review dataset. The result suggests that our model outperforms the state-of-the-art on semantic preservation, authorship obfuscation, and stylometric transformation.
2021-05-31
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (publié)
doi.org
arxiv.org
Explicitly Modeling Syntax in Language Models with Incremental Parsing and a Dynamic Oracle
Yikang Shen
Shawn Tan
Alessandro Sordoni
Siva Reddy
Aaron Courville
Syntax is fundamental to our thinking about language. Failing to capture the structure of input language could lead to generalization proble… (voir plus)ms and over-parametrization. In the present work, we propose a new syntax-aware language model: Syntactic Ordered Memory (SOM). The model explicitly models the structure with an incremental parser and maintains the conditional probability setting of a standard language model (left-to-right). To train the incremental parser and avoid exposure bias, we also propose a novel dynamic oracle, so that SOM is more robust to wrong parsing decisions. Experiments show that SOM can achieve strong results in language modeling, incremental parsing and syntactic generalization tests, while using fewer parameters than other models.
2021-05-31
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (publié)
doi.org
arxiv.org
Imperfect also Deserves Reward: Multi-Level and Sequential Reward Modeling for Better Dialog Management
Zhengxu Hou
Bang Liu
Ruihui Zhao
Zijing Ou
Yafei Liu
Xi Chen 0003
X. T. Chen
Yefeng Zheng
For task-oriented dialog systems, training a Reinforcement Learning (RL) based Dialog Management module suffers from low sample efficiency a… (voir plus)nd slow convergence speed due to the sparse rewards in RL. To solve this problem, many strategies have been proposed to give proper rewards when training RL, but their rewards lack interpretability and cannot accurately estimate the distribution of state-action pairs in real dialogs. In this paper, we propose a multi-level reward modeling approach that factorizes a reward into a three-level hierarchy: domain, act, and slot. Based on inverse adversarial reinforcement learning, our designed reward model can provide more accurate and explainable reward signals for state-action pairs. Extensive evaluations show that our approach can be applied to a wide range of reinforcement learning-based dialog systems and significantly improves both the performance and the speed of convergence.
2021-05-31
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (publié)
doi.org
arxiv.org
Improving Long-Term Metrics in Recommendation Systems using Short-Horizon Reinforcement Learning
Bogdan Mazoure
Paul Mineiro
Pavithra Srinath
Reza Sharifi Sedeh
Doina Precup
Adith Swaminathan
We study session-based recommendation scenarios where we want to recommend items to users during sequential interactions to improve their lo… (voir plus)ng-term utility. Optimizing a long-term metric is challenging because the learning signal (whether the recommendations achieved their desired goals) is delayed and confounded by other user interactions with the system. Targeting immediately measurable proxies such as clicks can lead to suboptimal recommendations due to misalignment with the long-term metric. We develop a new reinforcement learning algorithm called Short Horizon Policy Improvement (SHPI) that approximates policy-induced drift in user behavior across sessions. SHPI is a straightforward modification of episodic RL algorithms for session-based recommendation, that additionally gives an appropriate termination bonus in each session. Empirical results on four recommendation tasks show that SHPI can outperform state-of-the-art recommendation techniques like matrix factorization with offline proxy signals, bandits with myopic online proxies, and RL baselines with limited amounts of user interaction.
2021-05-31
ArXiv (prépublication)
arxiv.org
arxiv.org
Modeling Event Plausibility with Consistent Conceptual Abstraction
Ian Porada
Kaheer Suleman
Adam Trischler
Jackie CK Cheung
2021-05-31
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (publié)
doi.org
arxiv.org
Understanding by Understanding Not: Modeling Negation in Language Models
Arian Hosseini
Siva Reddy
Dzmitry Bahdanau
R Devon Hjelm
Alessandro Sordoni
Aaron Courville
Negation is a core construction in natural language. Despite being very successful on many tasks, state-of-the-art pre-trained language mode… (voir plus)ls often handle negation incorrectly. To improve language models in this regard, we propose to augment the language modeling objective with an unlikelihood objective that is based on negated generic sentences from a raw text corpus. By training BERT with the resulting combined objective we reduce the mean top~1 error rate to 4% on the negated LAMA dataset. We also see some improvements on the negated NLI benchmarks.
2021-05-31
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (publié)
doi.org
arxiv.org
Gotta Go Fast When Generating Data with Score-Based Models
Alexia Jolicoeur-Martineau
Fei Li
Rémi Piché-Taillefer
Tal Kachman
Ioannis Mitliagkas
Score-based (denoising diffusion) generative models have recently gained a lot of success in generating realistic and diverse data. These ap… (voir plus)proaches define a forward diffusion process for transforming data to noise and generate data by reversing it (thereby going from noise to data). Unfortunately, current score-based models generate data very slowly due to the sheer number of score network evaluations required by numerical SDE solvers. In this work, we aim to accelerate this process by devising a more efficient SDE solver. Existing approaches rely on the Euler-Maruyama (EM) solver, which uses a fixed step size. We found that naively replacing it with other SDE solvers fares poorly - they either result in low-quality samples or become slower than EM. To get around this issue, we carefully devise an SDE solver with adaptive step sizes tailored to score-based generative models piece by piece. Our solver requires only two score function evaluations, rarely rejects samples, and leads to high-quality samples. Our approach generates data 2 to 10 times faster than EM while achieving better or equal sample quality. For high-resolution images, our method leads to significantly higher quality samples than all other methods tested. Our SDE solver has the benefit of requiring no step size tuning.
2021-05-27
ArXiv (prépublication)
openreview.net
openreview.net
Noised Consistency Training for Text Summarization
J. Y. Liu
Qianren Mao
Bang Liu
Hao Peng
Hongdong Zhu
Jian-Xin Li
Neural abstractive summarization methods often require large quantities of labeled training data. However, labeling large amounts of summari… (voir plus)zation data is often prohibitive due to time, financial, and expertise constraints, which has limited the usefulness of summarization systems to practical applications. In this paper, we argue that this limitation can be overcome by a semi-supervised approach: consistency training which is to leverage large amounts of unlabeled data to improve the performance of supervised learning over a small corpus. The consistency regularization semi-supervised learning can regularize model predictions to be invariant to small noise applied to input articles. By adding noised unlabeled corpus to help regularize consistency training, this framework obtains comparative performance without using the full dataset. In particular, we have verified that leveraging large amounts of unlabeled data decently improves the performance of supervised learning over an insufficient labeled dataset.
2021-05-27
ArXiv (prépublication)
arxiv.org
arxiv.org
AndroidEnv: A Reinforcement Learning Platform for Android
Daniel Toyama
Philippe Hamel
Anita Gergely
Gheorghe Comanici
Amelia Glaese
Zafarali Ahmed
Tyler Jackson
Shibl Mourad
Doina Precup
We introduce AndroidEnv, an open-source platform for Reinforcement Learning (RL) research built on top of the Android ecosystem. AndroidEnv … (voir plus)allows RL agents to interact with a wide variety of apps and services commonly used by humans through a universal touchscreen interface. Since agents train on a realistic simulation of an Android device, they have the potential to be deployed on real devices. In this report, we give an overview of the environment, highlighting the significant features it provides for research, and we present an empirical evaluation of some popular reinforcement learning agents on a set of tasks built on this platform.
2021-05-26
ArXiv (prépublication)
arxiv.org
arxiv.org