Nous utilisons des témoins pour analyser le trafic et l’utilisation de notre site web, afin de personnaliser votre expérience. Vous pouvez désactiver ces technologies à tout moment, mais cela peut restreindre certaines fonctionnalités du site. Consultez notre Politique de protection de la vie privée pour en savoir plus.
Paramètre des cookies
Vous pouvez activer et désactiver les types de cookies que vous souhaitez accepter. Cependant certains choix que vous ferez pourraient affecter les services proposés sur nos sites (ex : suggestions, annonces personnalisées, etc.).
Cookies essentiels
Ces cookies sont nécessaires au fonctionnement du site et ne peuvent être désactivés. (Toujours actif)
Cookies analyse
Acceptez-vous l'utilisation de cookies pour mesurer l'audience de nos sites ?
Multimedia Player
Acceptez-vous l'utilisation de cookies pour afficher et vous permettre de regarder les contenus vidéo hébergés par nos partenaires (YouTube, etc.) ?
Publications
Fast reinforcement learning with generalized policy updates
The combination of reinforcement learning with deep learning is a promising approach to tackle important sequential decision-making problems… (voir plus) that are currently intractable. One obstacle to overcome is the amount of data needed by learning systems of this type. In this article, we propose to address this issue through a divide-and-conquer approach. We argue that complex decision problems can be naturally decomposed into multiple tasks that unfold in sequence or in parallel. By associating each task with a reward function, this problem decomposition can be seamlessly accommodated within the standard reinforcement-learning formalism. The specific way we do so is through a generalization of two fundamental operations in reinforcement learning: policy improvement and policy evaluation. The generalized version of these operations allow one to leverage the solution of some tasks to speed up the solution of others. If the reward function of a task can be well approximated as a linear combination of the reward functions of tasks previously solved, we can reduce a reinforcement-learning problem to a simpler linear regression. When this is not the case, the agent can still exploit the task solutions by using them to interact with and learn about the environment. Both strategies considerably reduce the amount of data needed to solve a reinforcement-learning problem.
2020-08-17
Proceedings of the National Academy of Sciences of the United States of America (publié)
Tensor Networks (TN) offer a powerful framework to efficiently represent very high-dimensional objects. TN have recently shown their potenti… (voir plus)al 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].
In this work we present a novel, robust transition generation technique that can serve as a new tool for 3D animators, based on adversarial … (voir plus)recurrent neural networks. The system synthesises high-quality motions that use temporally-sparse keyframes as animation constraints. This is reminiscent of the job of in-betweening in traditional animation pipelines, in which an animator draws motion frames between provided keyframes. We first show that a state-of-the-art motion prediction model cannot be easily converted into a robust transition generator when only adding conditioning information about future keyframes. To solve this problem, we then propose two novel additive embedding modifiers that are applied at each timestep to latent representations encoded inside the network's architecture. One modifier is a time-to-arrival embedding that allows variations of the transition length with a single model. The other is a scheduled target noise vector that allows the system to be robust to target distortions and to sample different transitions given fixed keyframes. To qualitatively evaluate our method, we present a custom MotionBuilder plugin that uses our trained model to perform in-betweening in production scenarios. To quantitatively evaluate performance on transitions and generalizations to longer time horizons, we present well-defined in-betweening benchmarks on a subset of the widely used Human3.6M dataset and on LaFAN1, a novel high quality motion capture dataset that is more appropriate for transition generation. We are releasing this new dataset along with this work, with accompanying code for reproducing our baseline results.
There is significant interest in using brain imaging data to predict non-brain-imaging phenotypes in individual participants. However, most … (voir plus)prediction studies are underpowered, relying on less than a few hundred participants, leading to low reliability and inflated prediction performance. Yet, small sample sizes are unavoidable when studying clinical populations or addressing focused neuroscience questions. Here, we propose a simple framework – “meta-matching” – to translate predictive models from large-scale datasets to new unseen non-brain-imaging phenotypes in boutique studies. The key observation is that many large-scale datasets collect a wide range inter-correlated phenotypic measures. Therefore, a unique phenotype from a boutique study likely correlates with (but is not the same as) some phenotypes in some large-scale datasets. Meta-matching exploits these correlations to boost prediction in the boutique study. We applied meta-matching to the problem of predicting non-brain-imaging phenotypes using resting-state functional connectivity (RSFC). Using the UK Biobank (N = 36,848), we demonstrated that meta-matching can boost the prediction of new phenotypes in small independent datasets by 100% to 400% in many scenarios. When considering relative prediction performance, meta-matching significantly improved phenotypic prediction even in samples with 10 participants. When considering absolute prediction performance, meta-matching significantly improved phenotypic prediction when there were least 50 participants. With a growing number of large-scale population-level datasets collecting an increasing number of phenotypic measures, our results represent a lower bound on the potential of meta-matching to elevate small-scale boutique studies.
The complexity of social interactions is a defining property of the human species. Many social neuroscience experiments have sought to map … (voir plus)perspective taking’, ‘empathy’, and other canonical psychological constructs to distinguishable brain circuits. This predominant research paradigm was seldom complemented by bottom-up studies of the unknown sources of variation that add up to measures of social brain structure; perhaps due to a lack of large population datasets. We aimed at a systematic de-construction of social brain morphology into its elementary building blocks in the UK Biobank cohort (n=~10,000). Coherent patterns of structural co-variation were explored within a recent atlas of social brain locations, enabled through translating autoencoder algorithms from deep learning. The artificial neural networks learned rich subnetwork representations that became apparent from social brain variation at population scale. The learned subnetworks carried essential information about the co-dependence configurations between social brain regions, with the nucleus accumbens, medial prefrontal cortex, and temporoparietal junction embedded at the core. Some of the uncovered subnetworks contributed to predicting examined social traits in general, while other subnetworks helped predict specific facets of social functioning, such as feelings of loneliness. Our population-level evidence indicates that hidden subsystems of the social brain underpin interindividual variation in dissociable aspects of social lifestyle.
We approach the problem of implicit regularization in deep learning from a geometrical viewpoint. We highlight a possible regularization eff… (voir plus)ect induced by a dynamical alignment of the neural tangent features introduced by Jacot et al, along a small number of task-relevant directions. By extrapolating a new analysis of Rademacher complexity bounds in linear models, we propose and study a new heuristic complexity measure for neural networks which captures this phenomenon, in terms of sequences of tangent kernel classes along in the learning trajectories.
