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
Multi-Task Reinforcement Learning as a Hidden-Parameter Block MDP
Multi-task reinforcement learning is a rich paradigm where information from previously seen environments can be leveraged for better perform… (voir plus)ance and improved sample-efficiency in new environments. In this work, we leverage ideas of common structure underlying a family of Markov decision processes (MDPs) to improve performance in the few-shot regime. We use assumptions of structure from Hidden-Parameter MDPs and Block MDPs to propose a new framework, HiP-BMDP, and approach for learning a common representation and universal dynamics model. To this end, we provide transfer and generalization bounds based on task and state similarity, along with sample complexity bounds that depend on the aggregate number of samples across tasks, rather than the number of tasks, a significant improvement over prior work. To demonstrate the efficacy of the proposed method, we empirically compare and show improvements against other multi-task and meta-reinforcement learning baselines.
Due to the realization that deep reinforcement learning algorithms trained on high-dimensional tasks can strongly overfit to their training … (voir plus)environments, there have been several studies that investigated the generalization performance of these algorithms. However, there has been no similar study that evaluated the generalization performance of algorithms that were specifically designed for generalization, i.e. meta-reinforcement learning algorithms. In this paper, we assess the generalization performance of these algorithms by leveraging high-dimensional, procedurally generated environments. We find that these algorithms can display strong overfitting when they are evaluated on challenging tasks. We also observe that scalability to high-dimensional tasks with sparse rewards remains a significant problem among many of the current meta-reinforcement learning algorithms. With these results, we highlight the need for developing meta-reinforcement learning algorithms that can both generalize and scale.
Training a deep neural network requires the model to go over training data for several epochs and update network parameters. In continual le… (voir plus)arning, this process results in catastrophic forgetting which is one of the core issues of this domain. Most proposed approaches for this issue try to compensate for the effects of parameter updates in the batch incremental setup in which the training model visits a lot of samples for several epochs. However, it is not realistic to expect training data will always be fed to model in a batch incremental setup. This paper proposes a chaotic stream learner that mimics the chaotic behavior of biological neurons and does not updates network parameters. In addition, it can work with fewer samples compared to deep learning models on stream learning setup. Our experiments on MNIST, CIFAR10, and Omniglot show that the chaotic stream learner has less catastrophic forgetting by its nature in comparison to a CNN model in continual learning.
The study of generalization of neural networks in gradient-based meta-learning has recently great research interest. Previous work on the st… (voir plus)udy of the objective landscapes within the scope of few-shot classification empirically demonstrated that generalization to new tasks might be linked to the average inner product between their respective gradients vectors (Guiroy et al., 2019). Following that work, we study the effect that meta-training has on the learned space of representation of the network. Notably, we demonstrate that the global similarity in the space of representation, measured by the average inner product between the embeddings of meta-test examples, also correlates to generalization. Based on these observations, we propose a novel model-selection criterion for gradient-based meta-learning and experimentally validate its effectiveness.
While deep reinforcement learning excels at solving tasks where large amounts of data can be collected through virtually unlimited interacti… (voir plus)on with the environment, learning from limited interaction remains a key challenge. We posit that an agent can learn more efficiently if we augment reward maximization with self-supervised objectives based on structure in its visual input and sequential interaction with the environment. Our method, Momentum Predictive Representations (MPR), trains an agent to predict its own latent state representations multiple steps into the future. We compute target representations for future states using an encoder which is an exponential moving average of the agent's parameters, and we make predictions using a learned transition model. On its own, this future prediction objective outperforms prior methods for sample-efficient deep RL from pixels. We further improve performance by adding data augmentation to the future prediction loss, which forces the agent's representations to be consistent across multiple views of an observation. Our full self-supervised objective, which combines future prediction and data augmentation, achieves a median human-normalized score of 0.444 on Atari in a setting limited to 100K steps of environment interaction, which is a 66% relative improvement over the previous state-of-the-art. Moreover, even in this limited data regime, MPR exceeds expert human scores on 6 out of 26 games.
Background Reward processing has been proposed to underpin atypical social behavior, a core feature of autism spectrum disorder (ASD). Howev… (voir plus)er, previous neuroimaging studies have yielded inconsistent results regarding the specificity of atypicalities for social rewards in ASD. Utilizing a large sample, we aimed to assess altered reward processing in response to reward type (social, monetary) and reward phase (anticipation, delivery) in ASD. Methods Functional magnetic resonance imaging during social and monetary reward anticipation and delivery was performed in 212 individuals with ASD (7.6-30.5 years) and 181 typically developing (TD) participants (7.6-30.8 years). Results Across social and monetary reward anticipation, whole-brain analyses (p0.05, family-wise error-corrected) showed hypoactivation of the right ventral striatum (VS) in ASD. Further, region of interest (ROI) analy
Genome-Wide Association Studies are typically conducted using linear models to find genetic variants associated with common diseases. In the… (voir plus)se studies, association testing is done on a variant-by-variant basis, possibly missing out on non-linear interaction effects between variants. Deep networks can be used to model these interactions, but they are difficult to train and interpret on large genetic datasets. We propose a method that uses the gradient based deep interpretability technique named DeepLIFT to show that known diabetes genetic risk factors can be identified using deep models along with possibly novel associations.
We present a multi-relational temporal Knowledge Graph based on the daily interactions between artifacts in GitHub, one of the largest socia… (voir plus)l coding platforms. Such representation enables posing many user-activity and project management questions as link prediction and time queries over the knowledge graph. In particular, we introduce two new datasets for i) interpolated time-conditioned link prediction and ii) extrapolated time-conditioned link/time prediction queries, each with distinguished properties. Our experiments on these datasets highlight the potential of adapting knowledge graphs to answer broad software engineering questions. Meanwhile, it also reveals the unsatisfactory performance of existing temporal models on extrapolated queries and time prediction queries in general. To overcome these shortcomings, we introduce an extension to current temporal models using relative temporal information with regards to past events.
