Maximum Entropy (MaxEnt) reinforcement learning is a powerful learning paradigm which seeks to maximize return under entropy regularization. However, action entropy does not necessarily coincide with state entropy, e.g., when multiple actions produce the same transition. Instead, we propose to maximize the transition entropy, i.e., the entropy of next states. We show that transition entropy can be described by two terms; namely, model-dependent transition entropy and action redundancy. Particularly, we explore the latter in both deterministic and stochastic settings and develop tractable approximation methods in a near model-free setup. We construct algorithms to minimize action redundancy and demonstrate their effectiveness on a synthetic environment with multiple redundant actions as well as contemporary benchmarks in Atari and Mujoco. Our results suggest that action redundancy is a fundamental problem in reinforcement learning.

Deep Reinforcement Learning (RL) methods rely on experience replay to approximate the minibatched supervised learning setting; however, unlike supervised learning where access to lots of training data is crucial to generalization, replay-based deep RL appears to struggle in the presence of extraneous data. Recent works have shown that the performance of Deep Q-Network (DQN) degrades when its replay memory becomes too large. This suggests that outdated experiences somehow impact the performance of deep RL, which should not be the case for off-policy methods like DQN. Consequently, we re-examine the motivation for sampling uniformly over a replay memory, and find that it may be flawed when using function approximation. We show that -- despite conventional wisdom -- sampling from the uniform distribution does not yield uncorrelated training samples and therefore biases gradients during training. Our theory prescribes a special non-uniform distribution to cancel this effect, and we propose a stratified sampling scheme to efficiently implement it.

Learning effective representations in image-based environments is crucial for sample efficient Reinforcement Learning (RL). Unfortunately, in RL, representation learning is confounded with the exploratory experience of the agent -- learning a useful representation requires diverse data, while effective exploration is only possible with coherent representations. Furthermore, we would like to learn representations that not only generalize across tasks but also accelerate downstream exploration for efficient task-specific training. To address these challenges we propose Proto-RL, a self-supervised framework that ties representation learning with exploration through prototypical representations. These prototypes simultaneously serve as a summarization of the exploratory experience of an agent as well as a basis for representing observations. We pre-train these task-agnostic representations and prototypes on environments without downstream task information. This enables state-of-the-art downstream policy learning on a set of difficult continuous control tasks.

In this paper, a proactive dynamic spectrum sharing scheme between 4G and 5G systems is proposed. In particular, a controller decides on the resource split between NR and LTE every subframe while accounting for future network states such as high interference subframes and multimedia broadcast single frequency network (MBSFN) subframes. To solve this problem, a deep reinforcement learning (RL) algorithm based on Monte Carlo Tree Search (MCTS) is proposed. The introduced deep RL architecture is trained offline whereby the controller predicts a sequence of future states of the wireless access network by simulating hypothetical bandwidth splits over time starting from the current network state. The action sequence resulting in the best reward is then assigned. This is realized by predicting the quantities most directly relevant to planning, i.e., the reward, the action probabilities, and the value for each network state. Simulation results show that the proposed scheme is able to take actions while accounting for future states instead of being greedy in each subframe. The results also show that the proposed framework improves system-level performance.

A key challenge for reinforcement learning is solving long-horizon planning and control problems. Recent work has proposed leveraging programs to help guide the learning algorithm in these settings. However, these approaches impose a high manual burden on the user since they must provide a guiding program for every new task they seek to achieve. We propose an approach that leverages program synthesis to automatically generate the guiding program. A key challenge is how to handle partially observable environments. We propose model predictive program synthesis, which trains a generative model to predict the unobserved portions of the world, and then synthesizes a program based on samples from this model in a way that is robust to its uncertainty. We evaluate our approach on a set of challenging benchmarks, including a 2D Minecraft-inspired ``craft'' environment where the agent must perform a complex sequence of subtasks to achieve its goal, a box-world environment that requires abstract reasoning, and a variant of the craft environment where the agent is a MuJoCo Ant. Our approach significantly outperforms several baselines, and performs essentially as well as an oracle that is given an effective program.

The two fields of machine learning and graphical causality arose and developed separately. However, there is now cross-pollination and increasing interest in both fields to benefit from the advances of the other. In the present paper, we review fundamental concepts of causal inference and relate them to crucial open problems of machine learning, including transfer and generalization, thereby assaying how causality can contribute to modern machine learning research. This also applies in the opposite direction: we note that most work in causality starts from the premise that the causal variables are given. A central problem for AI and causality is, thus, causal representation learning, the discovery of high-level causal variables from low-level observations. Finally, we delineate some implications of causality for machine learning and propose key research areas at the intersection of both communities.

In this paper we address the challenge of exploration in deep reinforcement learning for robotic manipulation tasks. In sparse goal settings, an agent does not receive any positive feedback until randomly achieving the goal, which becomes infeasible for longer control sequences. Inspired by touch-based exploration observed in children, we formulate an intrinsic reward based on the sum of forces between a robot's force sensors and manipulation objects that encourages physical interaction. Furthermore, we introduce contact-prioritized experience replay, a sampling scheme that prioritizes contact rich episodes and transitions. We show that our solution accelerates the exploration and outperforms state-of-the-art methods on three fundamental robot manipulation benchmarks.

Meta-learning for offline reinforcement learning (OMRL) is an understudied problem with tremendous potential impact by enabling RL algorithms in many real-world applications. A popular solution to the problem is to infer task identity as augmented state using a context-based encoder, for which efficient learning of task representations remains an open challenge. In this work, we improve upon one of the SOTA OMRL algorithms, FOCAL, by incorporating intra-task attention mechanism and inter-task contrastive learning objectives for more effective task inference and learning of control. Theoretical analysis and experiments are presented to demonstrate the superior performance, efficiency and robustness of our end-to-end and model free method compared to prior algorithms across multiple meta-RL benchmarks.

This paper presents a sensor-level mapless collision avoidance algorithm for use in mobile robots that map raw sensor data to linear and angular velocities and navigate in an unknown environment without a map. An efficient training strategy is proposed to allow a robot to learn from both human experience data and self-exploratory data. A game format simulation framework is designed to allow the human player to tele-operate the mobile robot to a goal and human action is also scored using the reward function. Both human player data and self-playing data are sampled using prioritized experience replay algorithm. The proposed algorithm and training strategy have been evaluated in two different experimental configurations: \textit{Environment 1}, a simulated cluttered environment, and \textit{Environment 2}, a simulated corridor environment, to investigate the performance. It was demonstrated that the proposed method achieved the same level of reward using only 16\% of the training steps required by the standard Deep Deterministic Policy Gradient (DDPG) method in Environment 1 and 20\% of that in Environment 2. In the evaluation of 20 random missions, the proposed method achieved no collision in less than 2~h and 2.5~h of training time in the two Gazebo environments respectively. The method also generated smoother trajectories than DDPG. The proposed method has also been implemented on a real robot in the real-world environment for performance evaluation. We can confirm that the trained model with the simulation software can be directly applied into the real-world scenario without further fine-tuning, further demonstrating its higher robustness than DDPG. The video and code are available: https://youtu.be/BmwxevgsdGc https://github.com/hanlinniu/turtlebot3_ddpg_collision_avoidance

Efficient point-to-point navigation in the presence of a background flow field is important for robotic applications such as ocean surveying. In such applications, robots may only have knowledge of their immediate surroundings or be faced with time-varying currents, which limits the use of optimal control techniques for planning trajectories. Here, we apply a novel Reinforcement Learning algorithm to discover time-efficient navigation policies to steer a fixed-speed swimmer through an unsteady two-dimensional flow field. The algorithm entails inputting environmental cues into a deep neural network that determines the swimmer's actions, and deploying Remember and Forget Experience replay. We find that the resulting swimmers successfully exploit the background flow to reach the target, but that this success depends on the type of sensed environmental cue. Surprisingly, a velocity sensing approach outperformed a bio-mimetic vorticity sensing approach by nearly two-fold in success rate. Equipped with local velocity measurements, the reinforcement learning algorithm achieved near 100% success in reaching the target locations while approaching the time-efficiency of paths found by a global optimal control planner.