Offline reinforcement learning--learning a policy from a batch of data--is known to be hard for general MDPs. These results motivate the need to look at specific classes of MDPs where offline reinforcement learning might be feasible. In this work, we explore a restricted class of MDPs to obtain guarantees for offline reinforcement learning. The key property, which we call Action Impact Regularity (AIR), is that actions primarily impact a part of the state (an endogenous component) and have limited impact on the remaining part of the state (an exogenous component). AIR is a strong assumption, but it nonetheless holds in a number of real-world domains including financial markets. We discuss algorithms that exploit the AIR property, and provide a theoretical analysis for an algorithm based on Fitted-Q Iteration. Finally, we demonstrate that the algorithm outperforms existing offline reinforcement learning algorithms across different data collection policies in simulated and real world environments where the regularity holds.

This two-part paper develops a paradigmatic theory and detailed methods of the joint electricity market design using reinforcement-learning (RL)-based simulation. In Part 2, this theory is further demonstrated by elaborating detailed methods of designing an electricity spot market (ESM), together with a reserved capacity product (RC) in the ancillary service market (ASM) and a virtual bidding (VB) product in the financial market (FM). Following the theory proposed in Part 1, firstly, market design options in the joint market are specified. Then, the Markov game model is developed, in which we show how to incorporate market design options and uncertain risks in model formulation. A multi-agent policy proximal optimization (MAPPO) algorithm is elaborated, as a practical implementation of the generalized market simulation method developed in Part 1. Finally, the case study demonstrates how to pick the best market design options by using some of the market operation performance indicators proposed in Part 1, based on the simulation results generated by implementing the MAPPO algorithm. The impacts of different market design options on market participants' bidding strategy preference are also discussed.

Reinforcement Learning (RL) algorithms are known to scale poorly to environments with many available actions, requiring numerous samples to learn an optimal policy. The traditional approach of considering the same fixed action space in every possible state implies that the agent must understand, while also learning to maximize its reward, to ignore irrelevant actions such as $\textit{inapplicable actions}$ (i.e. actions that have no effect on the environment when performed in a given state). Knowing this information can help reduce the sample complexity of RL algorithms by masking the inapplicable actions from the policy distribution to only explore actions relevant to finding an optimal policy. While this technique has been formalized for quite some time within the Automated Planning community with the concept of precondition in the STRIPS language, RL algorithms have never formally taken advantage of this information to prune the search space to explore. This is typically done in an ad-hoc manner with hand-crafted domain logic added to the RL algorithm. In this paper, we propose a more systematic approach to introduce this knowledge into the algorithm. We (i) standardize the way knowledge can be manually specified to the agent; and (ii) present a new framework to autonomously learn the partial action model encapsulating the precondition of an action jointly with the policy. We show experimentally that learning inapplicable actions greatly improves the sample efficiency of the algorithm by providing a reliable signal to mask out irrelevant actions. Moreover, we demonstrate that thanks to the transferability of the knowledge acquired, it can be reused in other tasks and domains to make the learning process more efficient.

In face of the pressing need of decarbonization in the power sector, the re-design of electricity market is necessary as a Marco-level approach to accommodate the high penetration of renewable generations, and to achieve power system operation security, economic efficiency, and environmental friendliness. However, existing market design methodologies suffer from the lack of coordination among energy spot market (ESM), ancillary service market (ASM) and financial market (FM), i.e., the "joint market", and the lack of reliable simulation-based verification. To tackle these deficiencies, this two-part paper develops a paradigmatic theory and detailed methods of the joint market design using reinforcement-learning (RL)-based simulation. In Part 1, the theory and framework of this novel market design philosophy are proposed. First, the controversial market design options while designing the joint market are summarized as the targeted research questions. Second, the Markov game model is developed to describe the bidding game in the joint market, incorporating the market design options to be determined. Third, a framework of deploying multiple types of RL algorithms to simulate the market model is developed. Finally, several market operation performance indicators are proposed to validate the market design based on the simulation results.

In this paper, we study a novel episodic risk-sensitive Reinforcement Learning (RL) problem, named Iterated CVaR RL, which aims to maximize the tail of the reward-to-go at each step, and focuses on tightly controlling the risk of getting into catastrophic situations at each stage. This formulation is applicable to real-world tasks that demand strong risk avoidance throughout the decision process, such as autonomous driving, clinical treatment planning and robotics. We investigate two performance metrics under Iterated CVaR RL, i.e., Regret Minimization and Best Policy Identification. For both metrics, we design efficient algorithms ICVaR-RM and ICVaR-BPI, respectively, and provide nearly matching upper and lower bounds with respect to the number of episodes $K$. We also investigate an interesting limiting case of Iterated CVaR RL, called Worst Path RL, where the objective becomes to maximize the minimum possible cumulative reward. For Worst Path RL, we propose an efficient algorithm with constant upper and lower bounds. Finally, our techniques for bounding the change of CVaR due to the value function shift and decomposing the regret via a distorted visitation distribution are novel, and can find applications in other risk-sensitive RL problems.

In cooperative multi-agent reinforcement learning (MARL), the environmental stochasticity and uncertainties will increase exponentially when the number of agents increases, which puts hard pressure on how to come up with a compact latent representation from partial observation for boosting value decomposition. To tackle these issues, we propose a simple yet powerful method that alleviates partial observability and efficiently promotes coordination by introducing the UNit-wise attentive State Representation (UNSR). In UNSR, each agent learns a compact and disentangled unit-wise state representation outputted from transformer blocks, and produces its local action-value function. The proposed UNSR is used to boost the value decomposition with a multi-head attention mechanism for producing efficient credit assignment in the mixing network, providing an efficient reasoning path between the individual value function and joint value function. Experimental results demonstrate that our method achieves superior performance and data efficiency compared to solid baselines on the StarCraft II micromanagement challenge. Additional ablation experiments also help identify the key factors contributing to the performance of UNSR.

Optimized trade execution is to sell (or buy) a given amount of assets in a given time with the lowest possible trading cost. Recently, reinforcement learning (RL) has been applied to optimized trade execution to learn smarter policies from market data. However, we find that many existing RL methods exhibit considerable overfitting which prevents them from real deployment. In this paper, we provide an extensive study on the overfitting problem in optimized trade execution. First, we model the optimized trade execution as offline RL with dynamic context (ORDC), where the context represents market variables that cannot be influenced by the trading policy and are collected in an offline manner. Under this framework, we derive the generalization bound and find that the overfitting issue is caused by large context space and limited context samples in the offline setting. Accordingly, we propose to learn compact representations for context to address the overfitting problem, either by leveraging prior knowledge or in an end-to-end manner. To evaluate our algorithms, we also implement a carefully designed simulator based on historical limit order book (LOB) data to provide a high-fidelity benchmark for different algorithms. Our experiments on the high-fidelity simulator demonstrate that our algorithms can effectively alleviate overfitting and achieve better performance.

Sample efficiency is a fundamental challenge in de novo molecular design. Ideally, molecular generative models should learn to satisfy a desired objective under minimal oracle evaluations (computational prediction or wet-lab experiment). This problem becomes more apparent when using oracles that can provide increased predictive accuracy but impose a significant cost. Consequently, these oracles cannot be directly optimized under a practical budget. Molecular generative models have shown remarkable sample efficiency when coupled with reinforcement learning, as demonstrated in the Practical Molecular Optimization (PMO) benchmark. Here, we propose a novel algorithm called Augmented Memory that combines data augmentation with experience replay. We show that scores obtained from oracle calls can be reused to update the model multiple times. We compare Augmented Memory to previously proposed algorithms and show significantly enhanced sample efficiency in an exploitation task and a drug discovery case study requiring both exploration and exploitation. Our method achieves a new state-of-the-art in the PMO benchmark which enforces a computational budget, outperforming the previous best performing method on 19/23 tasks.

We present a simple, sample-efficient algorithm for introducing large but directed learning steps in reinforcement learning (RL), through the use of evolutionary operators. The methodology uses a population of RL agents training with a common experience buffer, with occasional crossovers and mutations of the agents in order to search efficiently through the policy space. Unlike prior literature on combining evolutionary search (ES) with RL, this work does not generate a distribution of agents from a common mean and covariance matrix. Neither does it require the evaluation of the entire population of policies at every time step. Instead, we focus on gradient-based training throughout the life of every policy (individual), with a sparse amount of evolutionary exploration. The resulting algorithm is shown to be robust to hyperparameter variations. As a surprising corollary, we show that simply initialising and training multiple RL agents with a common memory (with no further evolutionary updates) outperforms several standard RL baselines.

In cooperative multi-agent reinforcement learning (MARL), where an agent coordinates with teammate(s) for a shared goal, it may sustain non-stationary caused by the policy change of teammates. Prior works mainly concentrate on the policy change during the training phase or teammates altering cross episodes, ignoring the fact that teammates may suffer from policy change suddenly within an episode, which might lead to miscoordination and poor performance as a result. We formulate the problem as an open Dec-POMDP, where we control some agents to coordinate with uncontrolled teammates, whose policies could be changed within one episode. Then we develop a new framework, fast teammates adaptation (Fastap), to address the problem. Concretely, we first train versatile teammates' policies and assign them to different clusters via the Chinese Restaurant Process (CRP). Then, we train the controlled agent(s) to coordinate with the sampled uncontrolled teammates by capturing their identifications as context for fast adaptation. Finally, each agent applies its local information to anticipate the teammates' context for decision-making accordingly. This process proceeds alternately, leading to a robust policy that can adapt to any teammates during the decentralized execution phase. We show in multiple multi-agent benchmarks that Fastap can achieve superior performance than multiple baselines in stationary and non-stationary scenarios.