Modern Reinforcement Learning (RL) algorithms are able to outperform humans in a wide variety of tasks. Multi-agent reinforcement learning (MARL) settings present additional challenges, and successful cooperation in mixed-motive groups of agents depends on a delicate balancing act between individual and group objectives. Social conventions and norms, often inspired by human institutions, are used as tools for striking this balance. In this paper, we examine a fundamental, well-studied social convention that underlies cooperation in both animal and human societies: dominance hierarchies. We adapt the ethological theory of dominance hierarchies to artificial agents, borrowing the established terminology and definitions with as few amendments as possible. We demonstrate that populations of RL agents, operating without explicit programming or intrinsic rewards, can invent, learn, enforce, and transmit a dominance hierarchy to new populations. The dominance hierarchies that emerge have a similar structure to those studied in chickens, mice, fish, and other species.
This study presents a novel approach to addressing offline reinforcement learning (RL) problems by reframing them as regression tasks that can be effectively solved using Decision Trees. Mainly, we introduce two distinct frameworks: return-conditioned and return-weighted decision tree policies (RCDTP and RWDTP), both of which achieve notable speed in agent training as well as inference, with training typically lasting less than a few minutes. Despite the simplification inherent in this reformulated approach to offline RL, our agents demonstrate performance that is at least on par with the established methods. We evaluate our methods on D4RL datasets for locomotion and manipulation, as well as other robotic tasks involving wheeled and flying robots. Additionally, we assess performance in delayed/sparse reward scenarios and highlight the explainability of these policies through action distribution and feature importance.
Evolutionary Reinforcement Learning (ERL), which integrates Evolutionary Algorithms (EAs) and Reinforcement Learning (RL) for optimization, has demonstrated remarkable performance advancements. By fusing both approaches, ERL has emerged as a promising research direction. This survey offers a comprehensive overview of the diverse research branches in ERL. Specifically, we systematically summarize recent advancements in related algorithms and identify three primary research directions: EA-assisted Optimization of RL, RL-assisted Optimization of EA, and synergistic optimization of EA and RL. Following that, we conduct an in-depth analysis of each research direction, organizing multiple research branches. We elucidate the problems that each branch aims to tackle and how the integration of EAs and RL addresses these challenges. In conclusion, we discuss potential challenges and prospective future research directions across various research directions. To facilitate researchers in delving into ERL, we organize the algorithms and codes involved on https://github.com/yeshenpy/Awesome-Evolutionary-Reinforcement-Learning.
Diversity plays a crucial role in improving the performance of multi-agent reinforcement learning (MARL). Currently, many diversity-based methods have been developed to overcome the drawbacks of excessive parameter sharing in traditional MARL. However, there remains a lack of a general metric to quantify policy differences among agents. Such a metric would not only facilitate the evaluation of the diversity evolution in multi-agent systems, but also provide guidance for the design of diversity-based MARL algorithms. In this paper, we propose the multi-agent policy distance (MAPD), a general tool for measuring policy differences in MARL. By learning the conditional representations of agents' decisions, MAPD can computes the policy distance between any pair of agents. Furthermore, we extend MAPD to a customizable version, which can quantify differences among agent policies on specified aspects. Based on the online deployment of MAPD, we design a multi-agent dynamic parameter sharing (MADPS) algorithm as an example of the MAPD's applications. Extensive experiments demonstrate that our method is effective in measuring differences in agent policies and specific behavioral tendencies. Moreover, in comparison to other methods of parameter sharing, MADPS exhibits superior performance.
Safe offline RL is a promising way to bypass risky online interactions towards safe policy learning. Most existing methods only enforce soft constraints, i.e., constraining safety violations in expectation below thresholds predetermined. This can lead to potentially unsafe outcomes, thus unacceptable in safety-critical scenarios. An alternative is to enforce the hard constraint of zero violation. However, this can be challenging in offline setting, as it needs to strike the right balance among three highly intricate and correlated aspects: safety constraint satisfaction, reward maximization, and behavior regularization imposed by offline datasets. Interestingly, we discover that via reachability analysis of safe-control theory, the hard safety constraint can be equivalently translated to identifying the largest feasible region given the offline dataset. This seamlessly converts the original trilogy problem to a feasibility-dependent objective, i.e., maximizing reward value within the feasible region while minimizing safety risks in the infeasible region. Inspired by these, we propose FISOR (FeasIbility-guided Safe Offline RL), which allows safety constraint adherence, reward maximization, and offline policy learning to be realized via three decoupled processes, while offering strong safety performance and stability. In FISOR, the optimal policy for the translated optimization problem can be derived in a special form of weighted behavior cloning. Thus, we propose a novel energy-guided diffusion model that does not require training a complicated time-dependent classifier to extract the policy, greatly simplifying the training. We compare FISOR against baselines on DSRL benchmark for safe offline RL. Evaluation results show that FISOR is the only method that can guarantee safety satisfaction in all tasks, while achieving top returns in most tasks.
We propose LangProp, a framework for iteratively optimizing code generated by large language models (LLMs), in both supervised and reinforcement learning settings. While LLMs can generate sensible coding solutions zero-shot, they are often sub-optimal. Especially for code generation tasks, it is likely that the initial code will fail on certain edge cases. LangProp automatically evaluates the code performance on a dataset of input-output pairs, catches any exceptions, and feeds the results back to the LLM in the training loop, so that the LLM can iteratively improve the code it generates. By adopting a metric- and data-driven training paradigm for this code optimization procedure, one could easily adapt findings from traditional machine learning techniques such as imitation learning, DAgger, and reinforcement learning. We show LangProp's applicability to general domains such as Sudoku and CartPole, as well as demonstrate the first proof of concept of automated code optimization for autonomous driving in CARLA. We show that LangProp can generate interpretable and transparent policies that can be verified and improved in a metric- and data-driven way. Our code is available at https://github.com/shuishida/LangProp.
We study the benefit of sharing representations among tasks to enable the effective use of deep neural networks in Multi-Task Reinforcement Learning. We leverage the assumption that learning from different tasks, sharing common properties, is helpful to generalize the knowledge of them resulting in a more effective feature extraction compared to learning a single task. Intuitively, the resulting set of features offers performance benefits when used by Reinforcement Learning algorithms. We prove this by providing theoretical guarantees that highlight the conditions for which is convenient to share representations among tasks, extending the well-known finite-time bounds of Approximate Value-Iteration to the multi-task setting. In addition, we complement our analysis by proposing multi-task extensions of three Reinforcement Learning algorithms that we empirically evaluate on widely used Reinforcement Learning benchmarks showing significant improvements over the single-task counterparts in terms of sample efficiency and performance.
A hallmark of intelligence is the ability to exhibit a wide range of effective behaviors. Inspired by this principle, Quality-Diversity algorithms, such as MAP-Elites, are evolutionary methods designed to generate a set of diverse and high-fitness solutions. However, as a genetic algorithm, MAP-Elites relies on random mutations, which can become inefficient in high-dimensional search spaces, thus limiting its scalability to more complex domains, such as learning to control agents directly from high-dimensional inputs. To address this limitation, advanced methods like PGA-MAP-Elites and DCG-MAP-Elites have been developed, which combine actor-critic techniques from Reinforcement Learning with MAP-Elites, significantly enhancing the performance and efficiency of Quality-Diversity algorithms in complex, high-dimensional tasks. While these methods have successfully leveraged the trained critic to guide more effective mutations, the potential of the trained actor remains underutilized in improving both the quality and diversity of the evolved population. In this work, we introduce DCRL-MAP-Elites, an extension of DCG-MAP-Elites that utilizes the descriptor-conditioned actor as a generative model to produce diverse solutions, which are then injected into the offspring batch at each generation. Additionally, we present an empirical analysis of the fitness and descriptor reproducibility of the solutions discovered by each algorithm. Finally, we present a second empirical analysis shedding light on the synergies between the different variations operators and explaining the performance improvement from PGA-MAP-Elites to DCRL-MAP-Elites.
Deploying controllers trained with Reinforcement Learning (RL) on real robots can be challenging: RL relies on agents' policies being modeled as Markov Decision Processes (MDPs), which assume an inherently discrete passage of time. The use of MDPs results in that nearly all RL-based control systems employ a fixed-rate control strategy with a period (or time step) typically chosen based on the developer's experience or specific characteristics of the application environment. Unfortunately, the system should be controlled at the highest, worst-case frequency to ensure stability, which can demand significant computational and energy resources and hinder the deployability of the controller on onboard hardware. Adhering to the principles of reactive programming, we surmise that applying control actions only when necessary enables the use of simpler hardware and helps reduce energy consumption. We challenge the fixed frequency assumption by proposing a variant of RL with variable control rate. In this approach, the policy decides the action the agent should take as well as the duration of the time step associated with that action. In our new setting, we expand Soft Actor-Critic (SAC) to compute the optimal policy with a variable control rate, introducing the Soft Elastic Actor-Critic (SEAC) algorithm. We show the efficacy of SEAC through a proof-of-concept simulation driving an agent with Newtonian kinematics. Our experiments show higher average returns, shorter task completion times, and reduced computational resources when compared to fixed rate policies.
Recent advances in Multi-Agent Reinforcement Learning have prompted the modeling of intricate interactions between agents in simulated environments. In particular, the predator-prey dynamics have captured substantial interest and various simulations been tailored to unique requirements. To prevent further time-intensive developments, we introduce Aquarium, a comprehensive Multi-Agent Reinforcement Learning environment for predator-prey interaction, enabling the study of emergent behavior. Aquarium is open source and offers a seamless integration of the PettingZoo framework, allowing a quick start with proven algorithm implementations. It features physics-based agent movement on a two-dimensional, edge-wrapping plane. The agent-environment interaction (observations, actions, rewards) and the environment settings (agent speed, prey reproduction, predator starvation, and others) are fully customizable. Besides a resource-efficient visualization, Aquarium supports to record video files, providing a visual comprehension of agent behavior. To demonstrate the environment's capabilities, we conduct preliminary studies which use PPO to train multiple prey agents to evade a predator. In accordance to the literature, we find Individual Learning to result in worse performance than Parameter Sharing, which significantly improves coordination and sample-efficiency.
De novo drug design is a pivotal issue in pharmacology and a new area of focus in AI for science research. A central challenge in this field is to generate molecules with specific properties while also producing a wide range of diverse candidates. Although advanced technologies such as transformer models and reinforcement learning have been applied in drug design, their potential has not been fully realized. Therefore, we propose MolRL-MGPT, a reinforcement learning algorithm with multiple GPT agents for drug molecular generation. To promote molecular diversity, we encourage the agents to collaborate in searching for desirable molecules in diverse directions. Our algorithm has shown promising results on the GuacaMol benchmark and exhibits efficacy in designing inhibitors against SARS-CoV-2 protein targets. The codes are available at: https://github.com/HXYfighter/MolRL-MGPT.
The availability of challenging benchmarks has played a key role in the recent progress of machine learning. In cooperative multi-agent reinforcement learning, the StarCraft Multi-Agent Challenge (SMAC) has become a popular testbed for centralised training with decentralised execution. However, after years of sustained improvement on SMAC, algorithms now achieve near-perfect performance. In this work, we conduct new analysis demonstrating that SMAC lacks the stochasticity and partial observability to require complex *closed-loop* policies. In particular, we show that an *open-loop* policy conditioned only on the timestep can achieve non-trivial win rates for many SMAC scenarios. To address this limitation, we introduce SMACv2, a new version of the benchmark where scenarios are procedurally generated and require agents to generalise to previously unseen settings (from the same distribution) during evaluation. We also introduce the extended partial observability challenge (EPO), which augments SMACv2 to ensure meaningful partial observability. We show that these changes ensure the benchmark requires the use of *closed-loop* policies. We evaluate state-of-the-art algorithms on SMACv2 and show that it presents significant challenges not present in the original benchmark. Our analysis illustrates that SMACv2 addresses the discovered deficiencies of SMAC and can help benchmark the next generation of MARL methods. Videos of training are available at https://sites.google.com/view/smacv2.
Personalization in machine learning (ML) tailors models' decisions to the individual characteristics of users. While this approach has seen success in areas like recommender systems, its expansion into high-stakes fields such as healthcare and autonomous driving is hindered by the extensive regulatory approval processes involved. To address this challenge, we propose a novel framework termed represented Markov Decision Processes (r-MDPs) that is designed to balance the need for personalization with the regulatory constraints. In an r-MDP, we cater to a diverse user population, each with unique preferences, through interaction with a small set of representative policies. Our objective is twofold: efficiently match each user to an appropriate representative policy and simultaneously optimize these policies to maximize overall social welfare. We develop two deep reinforcement learning algorithms that efficiently solve r-MDPs. These algorithms draw inspiration from the principles of classic K-means clustering and are underpinned by robust theoretical foundations. Our empirical investigations, conducted across a variety of simulated environments, showcase the algorithms' ability to facilitate meaningful personalization even under constrained policy budgets. Furthermore, they demonstrate scalability, efficiently adapting to larger policy budgets.
The reward signal plays a central role in defining the desired behaviors of agents in reinforcement learning (RL). Rewards collected from realistic environments could be perturbed, corrupted, or noisy due to an adversary, sensor error, or because they come from subjective human feedback. Thus, it is important to construct agents that can learn under such rewards. Existing methodologies for this problem make strong assumptions, including that the perturbation is known in advance, clean rewards are accessible, or that the perturbation preserves the optimal policy. We study a new, more general, class of unknown perturbations, and introduce a distributional reward critic framework for estimating reward distributions and perturbations during training. Our proposed methods are compatible with any RL algorithm. Despite their increased generality, we show that they achieve comparable or better rewards than existing methods in a variety of environments, including those with clean rewards. Under the challenging and generalized perturbations we study, we win/tie the highest return in 44/48 tested settings (compared to 11/48 for the best baseline). Our results broaden and deepen our ability to perform RL in reward-perturbed environments.
Radio Frequency Reinforcement Learning (RFRL) is anticipated to be a widely applicable technology in the next generation of wireless communication systems, particularly 6G and next-gen military communications. Given this, our research is focused on developing a tool to promote the development of RFRL techniques that leverage spectrum sensing. In particular, the tool was designed to address two cognitive radio applications, specifically dynamic spectrum access and jamming. In order to train and test reinforcement learning (RL) algorithms for these applications, a simulation environment is necessary to simulate the conditions that an agent will encounter within the Radio Frequency (RF) spectrum. In this paper, such an environment has been developed, herein referred to as the RFRL Gym. Through the RFRL Gym, users can design their own scenarios to model what an RL agent may encounter within the RF spectrum as well as experiment with different spectrum sensing techniques. Additionally, the RFRL Gym is a subclass of OpenAI gym, enabling the use of third-party ML/RL Libraries. We plan to open-source this codebase to enable other researchers to utilize the RFRL Gym to test their own scenarios and RL algorithms, ultimately leading to the advancement of RL research in the wireless communications domain. This paper describes in further detail the components of the Gym, results from example scenarios, and plans for future additions. Index Terms-machine learning, reinforcement learning, wireless communications, dynamic spectrum access, OpenAI gym
When deploying machine learning algorithms in the real world, guaranteeing safety is an essential asset. Existing safe learning approaches typically consider continuous variables, i.e., regression tasks. However, in practice, robotic systems are also subject to discrete, external environmental changes, e.g., having to carry objects of certain weights or operating on frozen, wet, or dry surfaces. Such influences can be modeled as discrete context variables. In the existing literature, such contexts are, if considered, mostly assumed to be known. In this work, we drop this assumption and show how we can perform safe learning when we cannot directly measure the context variables. To achieve this, we derive frequentist guarantees for multi-class classification, allowing us to estimate the current context from measurements. Further, we propose an approach for identifying contexts through experiments. We discuss under which conditions we can retain theoretical guarantees and demonstrate the applicability of our algorithm on a Furuta pendulum with camera measurements of different weights that serve as contexts.
Goal misalignment, reward sparsity and difficult credit assignment are only a few of the many issues that make it difficult for deep reinforcement learning (RL) agents to learn optimal policies. Unfortunately, the black-box nature of deep neural networks impedes the inclusion of domain experts for inspecting the model and revising suboptimal policies. To this end, we introduce *Successive Concept Bottleneck Agents* (SCoBots), that integrate consecutive concept bottleneck (CB) layers. In contrast to current CB models, SCoBots do not just represent concepts as properties of individual objects, but also as relations between objects which is crucial for many RL tasks. Our experimental results provide evidence of SCoBots' competitive performances, but also of their potential for domain experts to understand and regularize their behavior. Among other things, SCoBots enabled us to identify a previously unknown misalignment problem in the iconic video game, Pong, and resolve it. Overall, SCoBots thus result in more human-aligned RL agents. Our code is available at https://github.com/k4ntz/SCoBots .
Alleviating overestimation bias is a critical challenge for deep reinforcement learning to achieve successful performance on more complex tasks or offline datasets containing out-of-distribution data. In order to overcome overestimation bias, ensemble methods for Q-learning have been investigated to exploit the diversity of multiple Q-functions. Since network initialization has been the predominant approach to promote diversity in Q-functions, heuristically designed diversity injection methods have been studied in the literature. However, previous studies have not attempted to approach guaranteed independence over an ensemble from a theoretical perspective. By introducing a novel regularization loss for Q-ensemble independence based on random matrix theory, we propose spiked Wishart Q-ensemble independence regularization (SPQR) for reinforcement learning. Specifically, we modify the intractable hypothesis testing criterion for the Q-ensemble independence into a tractable KL divergence between the spectral distribution of the Q-ensemble and the target Wigner's semicircle distribution. We implement SPQR in several online and offline ensemble Q-learning algorithms. In the experiments, SPQR outperforms the baseline algorithms in both online and offline RL benchmarks.
In this paper, we aim to utilize only offline trajectory data to train a policy for multi-objective RL. We extend the offline policy-regularized method, a widely-adopted approach for single-objective offline RL problems, into the multi-objective setting in order to achieve the above goal. However, such methods face a new challenge in offline MORL settings, namely the preference-inconsistent demonstration problem. We propose two solutions to this problem: 1) filtering out preference-inconsistent demonstrations via approximating behavior preferences, and 2) adopting regularization techniques with high policy expressiveness. Moreover, we integrate the preference-conditioned scalarized update method into policy-regularized offline RL, in order to simultaneously learn a set of policies using a single policy network, thus reducing the computational cost induced by the training of a large number of individual policies for various preferences. Finally, we introduce Regularization Weight Adaptation to dynamically determine appropriate regularization weights for arbitrary target preferences during deployment. Empirical results on various multi-objective datasets demonstrate the capability of our approach in solving offline MORL problems.
Symbolic Music Alignment is the process of matching performed MIDI notes to corresponding score notes. In this paper, we introduce a reinforcement learning (RL)-based online symbolic music alignment technique. The RL agent - an attention-based neural network - iteratively estimates the current score position from local score and performance contexts. For this symbolic alignment task, environment states can be sampled exhaustively and the reward is dense, rendering a formulation as a simplified offline RL problem straightforward. We evaluate the trained agent in three ways. First, in its capacity to identify correct score positions for sampled test contexts; second, as the core technique of a complete algorithm for symbolic online note-wise alignment; and finally, as a real-time symbolic score follower. We further investigate the pitch-based score and performance representations used as the agent's inputs. To this end, we develop a second model, a two-step Dynamic Time Warping (DTW)-based offline alignment algorithm leveraging the same input representation. The proposed model outperforms a state-of-the-art reference model of offline symbolic music alignment.