Despite remarkable successes, deep reinforcement learning algorithms remain sample inefficient: they require an enormous amount of trial and error to find good policies. Model-based algorithms promise sample efficiency by building an environment model that can be used for planning. Posterior Sampling for Reinforcement Learning is such a model-based algorithm that has attracted significant interest due to its performance in the tabular setting. This paper introduces Posterior Sampling for Deep Reinforcement Learning (PSDRL), the first truly scalable approximation of Posterior Sampling for Reinforcement Learning that retains its model-based essence. PSDRL combines efficient uncertainty quantification over latent state space models with a specially tailored incremental planning algorithm based on value-function approximation. Extensive experiments on the Atari benchmark show that PSDRL significantly outperforms previous state-of-the-art attempts at scaling up posterior sampling while being competitive with a state-of-the-art (model-based) reinforcement learning method, both in sample efficiency and computational efficiency.

Meticulously analysing the empirical strengths and weaknesses of reinforcement learning methods in hard (challenging) environments is essential to inspire innovations and assess progress in the field. In tabular reinforcement learning, there is no well-established standard selection of environments to conduct such analysis, which is partially due to the lack of a widespread understanding of the rich theory of hardness of environments. The goal of this paper is to unlock the practical usefulness of this theory through four main contributions. First, we present a systematic survey of the theory of hardness, which also identifies promising research directions. Second, we introduce $\texttt{Colosseum}$, a pioneering package that enables empirical hardness analysis and implements a principled benchmark composed of environments that are diverse with respect to different measures of hardness. Third, we present an empirical analysis that provides new insights into computable measures. Finally, we benchmark five tabular agents in our newly proposed benchmark. While advancing the theoretical understanding of hardness in non-tabular reinforcement learning remains essential, our contributions in the tabular setting are intended as solid steps towards a principled non-tabular benchmark. Accordingly, we benchmark four agents in non-tabular versions of $\texttt{Colosseum}$ environments, obtaining results that demonstrate the generality of tabular hardness measures.

An agent in a nonstationary contextual bandit problem should balance between exploration and the exploitation of (periodic or structured) patterns present in its previous experiences. Handcrafting an appropriate historical context is an attractive alternative to transform a nonstationary problem into a stationary problem that can be solved efficiently. However, even a carefully designed historical context may introduce spurious relationships or lack a convenient representation of crucial information. In order to address these issues, we propose an approach that learns to represent the relevant context for a decision based solely on the raw history of interactions between the agent and the environment. This approach relies on a combination of features extracted by recurrent neural networks with a contextual linear bandit algorithm based on posterior sampling. Our experiments on a diverse selection of contextual and noncontextual nonstationary problems show that our recurrent approach consistently outperforms its feedforward counterpart, which requires handcrafted historical contexts, while being more widely applicable than conventional nonstationary bandit algorithms. Although it is very difficult to provide theoretical performance guarantees for our new approach, we also prove a novel regret bound for linear posterior sampling with measurement error that may serve as a foundation for future theoretical work.

A reinforcement learning agent that needs to pursue different goals across episodes requires a goal-conditional policy. In addition to their potential to generalize desirable behavior to unseen goals, such policies may also enable higher-level planning based on subgoals. In sparse-reward environments, the capacity to exploit information about the degree to which an arbitrary goal has been achieved while another goal was intended appears crucial to enabling sample efficient learning. However, reinforcement learning agents have only recently been endowed with such capacity for hindsight. In this letter, we demonstrate how hindsight can be introduced to policy gradient methods, generalizing this idea to a broad class of successful algorithms. Our experiments on a diverse selection of sparse-reward environments show that hindsight leads to a remarkable increase in sample efficiency.

Abstract We propose a new graph layout method based on a modification of the t‐distributed Stochastic Neighbor Embedding (t‐SNE) dimensionality reduction technique. Although t‐SNE is one of the best techniques for visualizing high‐dimensional data as 2D scatterplots, t‐SNE has not been used in the context of classical graph layout. We propose a new graph layout method, tsNET, based on representing a graph with a distance matrix, which together with a modified t‐SNE cost function results in desirable layouts. We evaluate our method by a formal comparison with state‐of‐the‐art methods, both visually and via established quality metrics on a comprehensive benchmark, containing real‐world and synthetic graphs. As evidenced by the quality metrics and visual inspection, tsNET produces excellent layouts.

In machine learning, pattern classification assigns high-dimensional vectors (observations) to classes based on generalization from examples. Artificial neural networks currently achieve state-of-the-art results in this task. Although such networks are typically used as black-boxes, they are also widely believed to learn (high-dimensional) higher-level representations of the original observations. In this paper, we propose using dimensionality reduction for two tasks: visualizing the relationships between learned representations of observations, and visualizing the relationships between artificial neurons. Through experiments conducted in three traditional image classification benchmark datasets, we show how visualization can provide highly valuable feedback for network designers. For instance, our discoveries in one of these datasets (SVHN) include the presence of interpretable clusters of learned representations, and the partitioning of artificial neurons into groups with apparently related discriminative roles.