Learning a predictive model of the mean return, or value function, plays a critical role in many reinforcement learning algorithms. Distributional reinforcement learning (DRL) methods instead model the value distribution, which has been shown to improve performance in many settings. In this paper, we model the value distribution as approximately normal using the Markov Chain central limit theorem. We analytically compute quantile bars to provide a new DRL target that is informed by the decrease in standard deviation that occurs over the course of an episode. In addition, we propose a policy update strategy based on uncertainty as measured by structural characteristics of the value distribution not present in the standard value function. The approach we outline is compatible with many DRL structures. We use two representative on-policy algorithms, PPO and TRPO, as testbeds and show that our methods produce performance improvements in continuous control tasks.

Proactive edge association is capable of improving wireless connectivity at the cost of increased handover (HO) frequency and energy consumption, while relying on a large amount of private information sharing required for decision making. In order to improve the connectivity-cost trade-off without privacy leakage, we investigate the privacy-preserving joint edge association and power allocation (JEAPA) problem in the face of the environmental uncertainty and the infeasibility of individual learning. Upon modelling the problem by a decentralized partially observable Markov Decision Process (Dec-POMDP), it is solved by federated multi-agent reinforcement learning (FMARL) through only sharing encrypted training data for federatively learning the policy sought. Our simulation results show that the proposed solution strikes a compelling trade-off, while preserving a higher privacy level than the state-of-the-art solutions.

Uncertainty in perception, actuation, and the environment often require multiple attempts for a robotic task to be successful. We study a class of problems providing (1) low-entropy indicators of terminal success / failure, and (2) unreliable (high-entropy) data to predict the final outcome of an ongoing task. Examples include a robot trying to connect with a charging station, parallel parking, or assembling a tightly-fitting part. The ability to restart after predicting failure early, versus simply running to failure, can significantly decrease the makespan, that is, the total time to completion, with the drawback of potentially short-cutting an otherwise successful operation. Assuming task running times to be Poisson distributed, and using a Markov Jump process to capture the dynamics of the underlying Markov Decision Process, we derive a closed form solution that predicts makespan based on the confusion matrix of the failure predictor. This allows the robot to learn failure prediction in a production environment, and only adopt a preemptive policy when it actually saves time. We demonstrate this approach using a robotic peg-in-hole assembly problem using a real robotic system. Failures are predicted by a dilated convolutional network based on force-torque data, showing an average makespan reduction from 101s to 81s (N=120, p<0.05). We posit that the proposed algorithm generalizes to any robotic behavior with an unambiguous terminal reward, with wide ranging applications on how robots can learn and improve their behaviors in the wild.

In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper we propose NeSIG, to the best of our knowledge the first domain-independent method for automatically generating planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on several classical domains, comparing our method with handcrafted domain-specific generators that generate valid and diverse problems but do not optimize difficulty. The results show NeSIG is able to automatically generate valid problems of greater difficulty than the competitor approaches, while maintaining good diversity.

Interpretability, explainability and transparency are key issues to introducing Artificial Intelligence methods in many critical domains: This is important due to ethical concerns and trust issues strongly connected to reliability, robustness, auditability and fairness, and has important consequences towards keeping the human in the loop in high levels of automation, especially in critical cases for decision making, where both (human and the machine) play important roles. While the research community has given much attention to explainability of closed (or black) prediction boxes, there are tremendous needs for explainability of closed-box methods that support agents to act autonomously in the real world. Reinforcement learning methods, and especially their deep versions, are such closed-box methods. In this article we aim to provide a review of state of the art methods for explainable deep reinforcement learning methods, taking also into account the needs of human operators - i.e., of those that take the actual and critical decisions in solving real-world problems. We provide a formal specification of the deep reinforcement learning explainability problems, and we identify the necessary components of a general explainable reinforcement learning framework. Based on these, we provide a comprehensive review of state of the art methods, categorizing them in classes according to the paradigm they follow, the interpretable models they use, and the surface representation of explanations provided. The article concludes identifying open questions and important challenges.

Methods for sequential decision-making are often built upon a foundational assumption that the underlying decision process is stationary. This limits the application of such methods because real-world problems are often subject to changes due to external factors (passive non-stationarity), changes induced by interactions with the system itself (active non-stationarity), or both (hybrid non-stationarity). In this work, we take the first steps towards the fundamental challenge of on-policy and off-policy evaluation amidst structured changes due to active, passive, or hybrid non-stationarity. Towards this goal, we make a higher-order stationarity assumption such that non-stationarity results in changes over time, but the way changes happen is fixed. We propose, OPEN, an algorithm that uses a double application of counterfactual reasoning and a novel importance-weighted instrument-variable regression to obtain both a lower bias and a lower variance estimate of the structure in the changes of a policy's past performances. Finally, we show promising results on how OPEN can be used to predict future performances for several domains inspired by real-world applications that exhibit non-stationarity.

Safety is a critical hurdle that limits the application of deep reinforcement learning (RL) to real-world control tasks. To this end, constrained reinforcement learning leverages cost functions to improve safety in constrained Markov decision processes. However, such constrained RL methods fail to achieve zero violation even when the cost limit is zero. This paper analyzes the reason for such failure, which suggests that a proper cost function plays an important role in constrained RL. Inspired by the analysis, we propose AutoCost, a simple yet effective framework that automatically searches for cost functions that help constrained RL to achieve zero-violation performance. We validate the proposed method and the searched cost function on the safe RL benchmark Safety Gym. We compare the performance of augmented agents that use our cost function to provide additive intrinsic costs with baseline agents that use the same policy learners but with only extrinsic costs. Results show that the converged policies with intrinsic costs in all environments achieve zero constraint violation and comparable performance with baselines.

We address the issue of safety in reinforcement learning. We pose the problem in an episodic framework of a constrained Markov decision process. Existing results have shown that it is possible to achieve a reward regret of $\tilde{\mathcal{O}}(\sqrt{K})$ while allowing an $\tilde{\mathcal{O}}(\sqrt{K})$ constraint violation in $K$ episodes. A critical question that arises is whether it is possible to keep the constraint violation even smaller. We show that when a strictly safe policy is known, then one can confine the system to zero constraint violation with arbitrarily high probability while keeping the reward regret of order $\tilde{\mathcal{O}}(\sqrt{K})$. The algorithm which does so employs the principle of optimistic pessimism in the face of uncertainty to achieve safe exploration. When no strictly safe policy is known, though one is known to exist, then it is possible to restrict the system to bounded constraint violation with arbitrarily high probability. This is shown to be realized by a primal-dual algorithm with an optimistic primal estimate and a pessimistic dual update.

Self-supervised pretraining has been extensively studied in language and vision domains, where a unified model can be easily adapted to various downstream tasks by pretraining representations without explicit labels. When it comes to sequential decision-making tasks, however, it is difficult to properly design such a pretraining approach that can cope with both high-dimensional perceptual information and the complexity of sequential control over long interaction horizons. The challenge becomes combinatorially more complex if we want to pretrain representations amenable to a large variety of tasks. To tackle this problem, in this work, we formulate a general pretraining-finetuning pipeline for sequential decision making, under which we propose a generic pretraining framework \textit{Self-supervised Multi-task pretrAining with contRol Transformer (SMART)}. By systematically investigating pretraining regimes, we carefully design a Control Transformer (CT) coupled with a novel control-centric pretraining objective in a self-supervised manner. SMART encourages the representation to capture the common essential information relevant to short-term control and long-term control, which is transferrable across tasks. We show by extensive experiments in DeepMind Control Suite that SMART significantly improves the learning efficiency among seen and unseen downstream tasks and domains under different learning scenarios including Imitation Learning (IL) and Reinforcement Learning (RL). Benefiting from the proposed control-centric objective, SMART is resilient to distribution shift between pretraining and finetuning, and even works well with low-quality pretraining datasets that are randomly collected.

The last decade, (2012 - 2022), saw an unprecedented advance in machine learning (ML) techniques, particularly deep learning (DL). As a result of the proven capabilities of DL, a large amount of work has been presented and studied in almost every field. Since 2012, when the convolution neural networks have been reintroduced in the context of \textit{ImagNet} competition, DL continued to achieve superior performance in many challenging tasks and problems. Wireless communications, in general, and medium access control (MAC) techniques, in particular, were among the fields that were heavily affected by this improvement. MAC protocols play a critical role in defining the performance of wireless communication systems. At the same time, the community lacks a comprehensive survey that collects, analyses, and categorizes the recent work in ML-inspired MAC techniques. In this work, we fill this gap by surveying a long line of work in this era. We solidify the impact of machine learning on wireless MAC protocols. We provide a comprehensive background to the widely adopted MAC techniques, their design issues, and their taxonomy, in connection with the famous application domains. Furthermore, we provide an overview of the ML techniques that have been considered in this context. Finally, we augment our work by proposing some promising future research directions and open research questions that are worth further investigation.