Designing a competent meta-reinforcement learning (meta-RL) algorithm in terms of data usage remains a central challenge to be tackled for its successful real-world applications. In this paper, we propose a sample-efficient meta-RL algorithm that learns a model of the system or environment at hand in a task-directed manner. As opposed to the standard model-based approaches to meta-RL, our method exploits the value information in order to rapidly capture the decision-critical part of the environment. The key component of our method is the loss function for learning the task inference module and the system model that systematically couples the model discrepancy and the value estimate, thereby facilitating the learning of the policy and the task inference module with a significantly smaller amount of data compared to the existing meta-RL algorithms. The idea is also extended to a non-meta-RL setting, namely an online linear quadratic regulator (LQR) problem, where our method can be simplified to reveal the essence of the strategy. The proposed method is evaluated in high-dimensional robotic control and online LQR problems, empirically verifying its effectiveness in extracting information indispensable for solving the tasks from observations in a sample efficient manner.

Informational parsimony provides a useful inductive bias for learning representations that achieve better generalization by being robust to noise and spurious correlations. We propose \textit{information gating} as a way to learn parsimonious representations that identify the minimal information required for a task. When gating information, we can learn to reveal as little information as possible so that a task remains solvable, or hide as little information as possible so that a task becomes unsolvable. We gate information using a differentiable parameterization of the signal-to-noise ratio, which can be applied to arbitrary values in a network, e.g., erasing pixels at the input layer or activations in some intermediate layer. When gating at the input layer, our models learn which visual cues matter for a given task. When gating intermediate layers, our models learn which activations are needed for subsequent stages of computation. We call our approach \textit{InfoGating}. We apply InfoGating to various objectives such as multi-step forward and inverse dynamics models, Q-learning, and behavior cloning, highlighting how InfoGating can naturally help in discarding information not relevant for control. Results show that learning to identify and use minimal information can improve generalization in downstream tasks. Policies based on InfoGating are considerably more robust to irrelevant visual features, leading to improved pretraining and finetuning of RL models.

Multi-relation question answering (QA) is a challenging task, where given questions usually require long reasoning chains in KGs that consist of multiple relations. Recently, methods with explicit multi-step reasoning over KGs have been prominently used in this task and have demonstrated promising performance. Examples include methods that perform stepwise label propagation through KG triples and methods that navigate over KG triples based on reinforcement learning. A main weakness of these methods is that their reasoning mechanisms are usually complex and difficult to implement or train. In this paper, we argue that multi-relation QA can be achieved via end-to-end single-step implicit reasoning, which is simpler, more efficient, and easier to adopt. We propose QAGCN -- a Question-Aware Graph Convolutional Network (GCN)-based method that includes a novel GCN architecture with controlled question-dependent message propagation for the implicit reasoning. Extensive experiments have been conducted, where QAGCN achieved competitive and even superior performance compared to state-of-the-art explicit-reasoning methods.

On-policy reinforcement learning (RL) methods seek to optimize a stochastic policy, where a neural network is used to parameterize a distribution π(a|s) over actions conditioned on the current state. In this framework, most on-policy RL methods seek to limit the scale of updates between successive policies during optimization. Some on-policy RL methods operate by guaranteeing that each policy update remains within a "trust region" (Schulman et al., 2015a). These methods are used when training stability during a long period of training is essential. However, finding a policy update near the edge of the trust region often comes at significant computational cost. Another branch of on-policy methods instead perform "proximal" policy updates, that limit the expected scale of policy updates, but can result in individual policy updates being of arbitrary magnitude (Schulman et al., 2017a). These methods are much more computationally efficient, but large-scale training can require the use of multiple training runs or human intervention to recover from training instabilities. In this work we propose Fixup Policy Optimization (FixPO), which combines both a proximal primary phase with a precise fixup phase, that operate by sharing a single penalty coefficient β. By performing a more conservative proximal update before strictly enforcing a trust region, FixPO is able to approximately match the computational efficiency and rewards of proximal methods while providing the same stability guarantees as trust region methods.

Neural Temporal Difference (TD) Learning is an approximate temporal difference method for policy evaluation that uses a neural network for function approximation. Analysis of Neural TD Learning has proven to be challenging. In this paper we provide a convergence analysis of Neural TD Learning with a projection onto $B(\theta_0, \omega)$, a ball of fixed radius $\omega$ around the initial point $\theta_0$. We show an approximation bound of $O(\epsilon) + \tilde{O} (1/\sqrt{m})$ where $\epsilon$ is the approximation quality of the best neural network in $B(\theta_0, \omega)$ and $m$ is the width of all hidden layers in the network.

Safe RL approaches are based on specific risk representations for particular problems or domains. In order to analyze agent behaviors, compare safe RL approaches, and effectively transfer techniques between application domains, it is necessary to understand the types of risk specific to safe RL problems. We performed a systematic literature mapping with the objective to characterize risk in safe RL. Based on the obtained results, we present definitions, characteristics, and types of risk that hold on multiple application domains. Our literature mapping covers literature from the last 5 years (2017-2022), from a variety of knowledge areas (AI, finance, engineering, medicine) where RL approaches emphasize risk representation and management. Our mapping covers 72 papers filtered systematically from over thousands of papers on the topic. Our proposed notion of risk covers a variety of representations, disciplinary differences, common training exercises, and types of techniques. We encourage researchers to include explicit and detailed accounts of risk in future safe RL research reports, using this mapping as a starting point. With this information, researchers and practitioners could draw stronger conclusions on the effectiveness of techniques on different problems.

Cooperation in multi-agent learning (MAL) is a topic at the intersection of numerous disciplines, including game theory, economics, social sciences, and evolutionary biology. Research in this area aims to understand both how agents can coordinate effectively when goals are aligned and how they may cooperate in settings where gains from working together are possible but possibilities for conflict abound. In this paper we provide an overview of the fundamental concepts, problem settings and algorithms of multi-agent learning. This encompasses reinforcement learning, multi-agent sequential decision-making, challenges associated with multi-agent cooperation, and a comprehensive review of recent progress, along with an evaluation of relevant metrics. Finally we discuss open challenges in the field with the aim of inspiring new avenues for research.

Ford Motor China Co.,ltd, Shanghai, 200082, China Abstract The comprehensiveness of vehicle-to-everything (V2X) recognition enriches and holistically shapes the global Birds-Eye-View (BEV) perception, incorporating rich semantics and integrating driving scene information, thereby serving features of trajectory prediction, decision-making and driving planning. Utilizing V2X message sets to form BEV format proves to be an effective perception method for connected and automated vehicles (CAVs). Specifically, MAP, SPAT and RSI data contributes to the achievement of road connectivity, synchronized traffic signal navigation and obstacle warning. Moreover, using time-sequential BSMs information from multiple vehicles allows for the perception of current state and the prediction of future trajectories. Therefore, this paper develops a comprehensive autonomous driving model that relies on BEV-V2X perception, Interacting Multiple model Unscented Kalman Filter (IMM-UKF)-based trajectory prediction, and deep reinforcement learning (DRL)-based decision making and planing. We establish a DRL environment with reward-shaping methods to formulate a unified set of optimal driving behaviors that encompass obstacle avoidance, lane changes, overtaking, turning maneuver, and synchronized traffic signal navigation. Consequently, a complex traffic intersection scenario was simulated, and the well-trained model was applied for driving control. The observed driving behavior closely resembled that of an experienced driver, exhibiting anticipatory actions and revealing notable operational highlights of driving policy. Ford Motor China Co.,ltd, Shanghai, 200082, China I. INTRODUCTION Cooperative Intelligent Transport Systems (C-ITS) based on vehicle-to-everything (V2X) communication has considerably developed in recent years, bringing us closer to utilizing V2X for autonomous driving and assistance guidance [1-2]. Due to the numerous substantial benefits provided by C-ITS, governmental authorities worldwide have initiated the allocation of dedicated spectrum for V2X technologies, on a license-exempt basis in the 5.9 GHz band for the cellular based Cellular-V2X (C-V2X) PC5 technology [3-4]. The characteristics of 5G technologies including enhanced Mobile Broadband (eMBB) and massive Machine Type Communications (mMTC) are pivotal in vehicular communications [5-6]. Furthermore, facilitating Ultra-Reliable and Low-Latency Communications (URLLC) of 5G is a fundamental element of advanced V2X applications [7-8].

Traffic congestion is a persistent problem in urban areas, which calls for the development of effective traffic signal control (TSC) systems. While existing Reinforcement Learning (RL)-based methods have shown promising performance in optimizing TSC, it is challenging to generalize these methods across intersections of different structures. In this work, a universal RL-based TSC framework is proposed for Vehicle-to-Everything (V2X) environments. The proposed framework introduces a novel agent design that incorporates a junction matrix to characterize intersection states, making the proposed model applicable to diverse intersections. To equip the proposed RL-based framework with enhanced capability of handling various intersection structures, novel traffic state augmentation methods are tailor-made for signal light control systems. Finally, extensive experimental results derived from multiple intersection configurations confirm the effectiveness of the proposed framework. The source code in this work is available at https://github.com/wmn7/Universal_Light

This paper proposes a new control framework for manipulating soft objects. A Deep Reinforcement Learning (DRL) approach is used to make the shape of a deformable object reach a set of desired points by controlling a robotic arm which manipulates it. Our framework is more easily generalizable than existing ones: it can work directly with different initial and desired final shapes without need for relearning. We achieve this by using learning parallelization, i.e., executing multiple agents in parallel on various environment instances. We focus our study on deformable linear objects. These objects are interesting in industrial and agricultural domains, yet their manipulation with robots, especially in 3D workspaces, remains challenging. We simulate the entire environment, i.e., the soft object and the robot, for the training and the testing using PyBullet and OpenAI Gym. We use a combination of state-of-the-art DRL techniques, the main ingredient being a training approach for the learning agent (i.e., the robot) based on Deep Deterministic Policy Gradient (DDPG). Our simulation results support the usefulness and enhanced generality of the proposed approach.