Publications

Compositional Attention: Disentangling Search and Retrieval

Multi-head, key-value attention is the backbone of the widely successful Transformer model and its variants. This attention mechanism uses m… (voir plus)ultiple parallel key-value attention blocks (called heads), each performing two fundamental computations: (1) search - selection of a relevant entity from a set via query-key interactions, and (2) retrieval - extraction of relevant features from the selected entity via a value matrix. Importantly, standard attention heads learn a rigid mapping between search and retrieval. In this work, we first highlight how this static nature of the pairing can potentially: (a) lead to learning of redundant parameters in certain tasks, and (b) hinder generalization. To alleviate this problem, we propose a novel attention mechanism, called Compositional Attention, that replaces the standard head structure. The proposed mechanism disentangles search and retrieval and composes them in a dynamic, flexible and context-dependent manner through an additional soft competition stage between the query-key combination and value pairing. Through a series of numerical experiments, we show that it outperforms standard multi-head attention on a variety of tasks, including some out-of-distribution settings. Through our qualitative analysis, we demonstrate that Compositional Attention leads to dynamic specialization based on the type of retrieval needed. Our proposed mechanism generalizes multi-head attention, allows independent scaling of search and retrieval, and can easily be implemented in lieu of standard attention heads in any network architecture.

2022-04-24

International Conference on Learning Representations (Accept (Spotlight))

doi.org

openreview.net

Constructing a Good Behavior Basis for Transfer Using Generalized Policy Updates

Safa Alver

Doina Precup

We study the problem of learning a good set of policies, so that when combined together, they can solve a wide variety of unseen reinforceme… (voir plus)nt learning tasks with no or very little new data. Specifically, we consider the framework of generalized policy evaluation and improvement, in which the rewards for all tasks of interest are assumed to be expressible as a linear combination of a fixed set of features. We show theoretically that, under certain assumptions, having access to a specific set of diverse policies, which we call a set of independent policies, can allow for instantaneously achieving high-level performance on all possible downstream tasks which are typically more complex than the ones on which the agent was trained. Based on this theoretical analysis, we propose a simple algorithm that iteratively constructs this set of policies. In addition to empirically validating our theoretical results, we compare our approach with recently proposed diverse policy set construction methods and show that, while others fail, our approach is able to build a behavior basis that enables instantaneous transfer to all possible downstream tasks. We also show empirically that having access to a set of independent policies can better bootstrap the learning process on downstream tasks where the new reward function cannot be described as a linear combination of the features. Finally, we demonstrate how this policy set can be useful in a lifelong reinforcement learning setting.

2022-04-24

International Conference on Learning Representations (Accept (Poster))

doi.org

openreview.net

Continuous-Time Meta-Learning with Forward Mode Differentiation

Drawing inspiration from gradient-based meta-learning methods with infinitely small gradient steps, we introduce Continuous-Time Meta-Learni… (voir plus)ng (COMLN), a meta-learning algorithm where adaptation follows the dynamics of a gradient vector field. Specifically, representations of the inputs are meta-learned such that a task-specific linear classifier is obtained as a solution of an ordinary differential equation (ODE). Treating the learning process as an ODE offers the notable advantage that the length of the trajectory is now continuous, as opposed to a fixed and discrete number of gradient steps. As a consequence, we can optimize the amount of adaptation necessary to solve a new task using stochastic gradient descent, in addition to learning the initial conditions as is standard practice in gradient-based meta-learning. Importantly, in order to compute the exact meta-gradients required for the outer-loop updates, we devise an efficient algorithm based on forward mode differentiation, whose memory requirements do not scale with the length of the learning trajectory, thus allowing longer adaptation in constant memory. We provide analytical guarantees for the stability of COMLN, we show empirically its efficiency in terms of runtime and memory usage, and we illustrate its effectiveness on a range of few-shot image classification problems.

2022-04-24

ICLR (Accept (Spotlight))

doi.org

openreview.net

Coordination Among Neural Modules Through a Shared Global Workspace

Nan Rosemary Ke

Nasim Rahaman

Jonathan Binas

Charles Blundell

Michael Mozer

Yoshua Bengio

Deep learning has seen a movement away from representing examples with a monolithic hidden state towards a richly structured state. For exam… (voir plus)ple, Transformers segment by position, and object-centric architectures decompose images into entities. In all these architectures, interactions between different elements are modeled via pairwise interactions: Transformers make use of self-attention to incorporate information from other positions; object-centric architectures make use of graph neural networks to model interactions among entities. However, pairwise interactions may not achieve global coordination or a coherent, integrated representation that can be used for downstream tasks. In cognitive science, a global workspace architecture has been proposed in which functionally specialized components share information through a common, bandwidth-limited communication channel. We explore the use of such a communication channel in the context of deep learning for modeling the structure of complex environments. The proposed method includes a shared workspace through which communication among different specialist modules takes place but due to limits on the communication bandwidth, specialist modules must compete for access. We show that capacity limitations have a rational basis in that (1) they encourage specialization and compositionality and (2) they facilitate the synchronization of otherwise independent specialists.

2022-04-24

ICLR (Accept (Oral))

doi.org

openreview.net

Exposing Query Identification for Search Transparency

Ruohan Li

Jianxiang Li

Bhaskar Mitra

Fernando Diaz

Asia J. Biega

Search systems control the exposure of ranked content to searchers. In many cases, creators value not only the exposure of their content but… (voir plus), moreover, an understanding of the specific searches where the content is surfaced. The problem of identifying which queries expose a given piece of content in the ranking results is an important and relatively under-explored search transparency challenge. Exposing queries are useful for quantifying various issues of search bias, privacy, data protection, security, and search engine optimization. Exact identification of exposing queries in a given system is computationally expensive, especially in dynamic contexts such as web search. We explore the feasibility of approximate exposing query identification (EQI) as a retrieval task by reversing the role of queries and documents in two classes of search systems: dense dual-encoder models and traditional BM25 models. We then propose how this approach can be improved through metric learning over the retrieval embedding space. We further derive an evaluation metric to measure the quality of a ranking of exposing queries, as well as conducting an empirical analysis focusing on various practical aspects of approximate EQI. Overall, our work contributes a novel conception of transparency in search systems and computational means of achieving it.

2022-04-24

ACM Web Conference (publié)

doi.org

arxiv.org

Generalized Natural Gradient Flows in Hidden Convex-Concave Games and Gans

Andjela Mladenovic

Iosif Sakos

Gauthier Gidel

Georgios Piliouras

2022-04-24

International Conference on Learning Representations (Accept (Poster))

openreview.net

Graph Neural Networks with Learnable Structural and Positional Representations

Vijay Prakash Dwivedi

Anh Tuan Luu

Thomas Laurent

Yoshua Bengio

Xavier Bresson

Graph neural networks (GNNs) have become the standard learning architectures for graphs. GNNs have been applied to numerous domains ranging … (voir plus)from quantum chemistry, recommender systems to knowledge graphs and natural language processing. A major issue with arbitrary graphs is the absence of canonical positional information of nodes, which decreases the representation power of GNNs to distinguish e.g. isomorphic nodes and other graph symmetries. An approach to tackle this issue is to introduce Positional Encoding (PE) of nodes, and inject it into the input layer, like in Transformers. Possible graph PE are Laplacian eigenvectors. In this work, we propose to decouple structural and positional representations to make easy for the network to learn these two essential properties. We introduce a novel generic architecture which we call LSPE (Learnable Structural and Positional Encodings). We investigate several sparse and fully-connected (Transformer-like) GNNs, and observe a performance increase for molecular datasets, from 1.79% up to 64.14% when considering learnable PE for both GNN classes.

2022-04-24

International Conference on Learning Representations (Accept (Poster))

doi.org

openreview.net

Latent Variable Sequential Set Transformers for Joint Multi-agent Motion Prediction

Jim Aldon D’Souza

Samira E. Kahou

Felix Heide

Christopher Pal

Robust multi-agent trajectory prediction is essential for the safe control of robotic systems. A major challenge is to efficiently learn a r… (voir plus)epresentation that approximates the true joint distribution of contextual, social, and temporal information to enable planning. We propose Latent Variable Sequential Set Transformers which are encoder-decoder architectures that generate scene-consistent multi-agent trajectories. We refer to these architectures as "AutoBots". The encoder is a stack of interleaved temporal and social multi-head self-attention (MHSA) modules which alternately perform equivariant processing across the temporal and social dimensions. The decoder employs learnable seed parameters in combination with temporal and social MHSA modules allowing it to perform inference over the entire future scene in a single forward pass efficiently. AutoBots can produce either the trajectory of one ego-agent or a distribution over the future trajectories for all agents in the scene. For the single-agent prediction case, our model achieves top results on the global nuScenes vehicle motion prediction leaderboard, and produces strong results on the Argoverse vehicle prediction challenge. In the multi-agent setting, we evaluate on the synthetic partition of TrajNet++ dataset to showcase the model's socially-consistent predictions. We also demonstrate our model on general sequences of sets and provide illustrative experiments modelling the sequential structure of the multiple strokes that make up symbols in the Omniglot data. A distinguishing feature of AutoBots is that all models are trainable on a single desktop GPU (1080 Ti) in under 48h.

2022-04-24

International Conference on Learning Representations (Accept (Spotlight))

doi.org

openreview.net

Learning to Guide and to Be Guided in the Architect-Builder Problem

Paul Barde

Tristan Karch

Derek Nowrouzezahrai

Clément Moulin-Frier

Christopher Pal

Pierre-Yves Oudeyer

We are interested in interactive agents that learn to coordinate, namely, a …

2022-04-24

International Conference on Learning Representations (Accept (Poster))

doi.org

openreview.net

Memory Augmented Optimizers for Deep Learning

Paul-Aymeric McRae

Prasanna Parthasarathi

Mahmoud Assran

Sarath Chandar

Popular approaches for minimizing loss in data-driven learning often involve an abstraction or an explicit retention of the history of gradi… (voir plus)ents for efficient parameter updates. The aggregated history of gradients nudges the parameter updates in the right direction even when the gradients at any given step are not informative. Although the history of gradients summarized in meta-parameters or explicitly stored in memory has been shown effective in theory and practice, the question of whether

2022-04-24

International Conference on Learning Representations (Accept (Poster))

doi.org

openreview.net

Meta-matching as a simple framework to translate phenotypic predictive models from big to small data

Tong He

Lijun An

Pansheng Chen

Jianzhong Chen

Jiashi Feng

Danilo Bzdok

Avram J. Holmes

Simon B. Eickhoff

B. T. Thomas Yeo

We propose a simple framework—meta-matching—to translate predictive models from large-scale datasets to new unseen non-brain-imaging phe… (voir plus)notypes in small-scale studies. The key consideration is that a unique phenotype from a boutique study likely correlates with (but is not the same as) related phenotypes in some large-scale dataset. Meta-matching exploits these correlations to boost prediction in the boutique study. We apply meta-matching to predict non-brain-imaging phenotypes from resting-state functional connectivity. Using the UK Biobank (N = 36,848) and Human Connectome Project (HCP) (N = 1,019) datasets, we demonstrate that meta-matching can greatly boost the prediction of new phenotypes in small independent datasets in many scenarios. For example, translating a UK Biobank model to 100 HCP participants yields an eight-fold improvement in variance explained with an average absolute gain of 4.0% (minimum = −0.2%, maximum = 16.0%) across 35 phenotypes. With a growing number of large-scale datasets collecting increasingly diverse phenotypes, our results represent a lower bound on the potential of meta-matching. Individual-level prediction is critical for precision medicine, but many neuroimaging prediction studies are underpowered. Here the authors present a simple yet powerful approach that effectively translates predictive models from big to small data.

2022-04-24

Nature Neuroscience (publié)

doi.org

MIDI-DDSP: Detailed Control of Musical Performance via Hierarchical Modeling

Yusong Wu

Ethan Manilow

Yi Deng

Rigel Swavely

Kyle Kastner

Tim Cooijmans

Aaron Courville

Cheng-Zhi Anna Huang

Jesse Engel

Musical expression requires control of both what notes are played, and how they are performed. Conventional audio synthesizers provide detai… (voir plus)led expressive controls, but at the cost of realism. Black-box neural audio synthesis and concatenative samplers can produce realistic audio, but have few mechanisms for control. In this work, we introduce MIDI-DDSP a hierarchical model of musical instruments that enables both realistic neural audio synthesis and detailed user control. Starting from interpretable Differentiable Digital Signal Processing (DDSP) synthesis parameters, we infer musical notes and high-level properties of their expressive performance (such as timbre, vibrato, dynamics, and articulation). This creates a 3-level hierarchy (notes, performance, synthesis) that affords individuals the option to intervene at each level, or utilize trained priors (performance given notes, synthesis given performance) for creative assistance. Through quantitative experiments and listening tests, we demonstrate that this hierarchy can reconstruct high-fidelity audio, accurately predict performance attributes for a note sequence, independently manipulate the attributes of a given performance, and as a complete system, generate realistic audio from a novel note sequence. By utilizing an interpretable hierarchy, with multiple levels of granularity, MIDI-DDSP opens the door to assistive tools to empower individuals across a diverse range of musical experience.

2022-04-24

International Conference on Learning Representations (Accept (Oral))

doi.org

openreview.net

Formation d'entrepreneurs scientifiques

Boussole des politiques en IA

Demandes de supervision

Publications

Formation d'entrepreneurs scientifiques

Boussole des politiques en IA

Demandes de supervision

Mots-clés populaires:

Publications