Reinforcement Learning (RL) is used extensively in Autonomous Systems (AS) as it enables learning at runtime without the need for a model of the environment or predefined actions. However, most applications of RL in AS, such as those based on Q-learning, can only optimize one objective, making it necessary in multi-objective systems to combine multiple objectives in a single objective function with predefined weights. A number of Multi-Objective Reinforcement Learning (MORL) techniques exist but they have mostly been applied in RL benchmarks rather than real-world AS systems. In this work, we use a MORL technique called Deep W-Learning (DWN) and apply it to the Emergent Web Servers exemplar, a self-adaptive server, to find the optimal configuration for runtime performance optimization. We compare DWN to two single-objective optimization implementations: {\epsilon}-greedy algorithm and Deep Q-Networks. Our initial evaluation shows that DWN optimizes multiple objectives simultaneously with similar results than DQN and {\epsilon}-greedy approaches, having a better performance for some metrics, and avoids issues associated with combining multiple objectives into a single utility function.
Self-play, characterized by agents' interactions with copies or past versions of themselves, has recently gained prominence in reinforcement learning (RL). This paper first clarifies the preliminaries of self-play, including the multi-agent reinforcement learning framework and basic game theory concepts. Then, it provides a unified framework and classifies existing self-play algorithms within this framework. Moreover, the paper bridges the gap between the algorithms and their practical implications by illustrating the role of self-play in different scenarios. Finally, the survey highlights open challenges and future research directions in self-play. This paper is an essential guide map for understanding the multifaceted landscape of self-play in RL.
When faced with a new customer, many factors contribute to an insurance firm's decision of what offer to make to that customer. In addition to the expected cost of providing the insurance, the firm must consider the other offers likely to be made to the customer, and how sensitive the customer is to differences in price. Moreover, firms often target a specific portfolio of customers that could depend on, e.g., age, location, and occupation. Given such a target portfolio, firms may choose to modulate an individual customer's offer based on whether the firm desires the customer within their portfolio. We term the problem of modulating offers to achieve a desired target portfolio the portfolio pursuit problem. Having formulated the portfolio pursuit problem as a sequential decision making problem, we devise a novel reinforcement learning algorithm for its solution. We test our method on a complex synthetic market environment, and demonstrate that it outperforms a baseline method which mimics current industry approaches to portfolio pursuit.
For on-policy reinforcement learning, discretizing action space for continuous control can easily express multiple modes and is straightforward to optimize. However, without considering the inherent ordering between the discrete atomic actions, the explosion in the number of discrete actions can possess undesired properties and induce a higher variance for the policy gradient estimator. In this paper, we introduce a straightforward architecture that addresses this issue by constraining the discrete policy to be unimodal using Poisson probability distributions. This unimodal architecture can better leverage the continuity in the underlying continuous action space using explicit unimodal probability distributions. We conduct extensive experiments to show that the discrete policy with the unimodal probability distribution provides significantly faster convergence and higher performance for on-policy reinforcement learning algorithms in challenging control tasks, especially in highly complex tasks such as Humanoid. We provide theoretical analysis on the variance of the policy gradient estimator, which suggests that our attentively designed unimodal discrete policy can retain a lower variance and yield a stable learning process.
Reinforcement Learning (RL) agents demonstrating proficiency in a training environment exhibit vulnerability to adversarial perturbations in input observations during deployment. This underscores the importance of building a robust agent before its real-world deployment. To alleviate the challenging point, prior works focus on developing robust training-based procedures, encompassing efforts to fortify the deep neural network component's robustness or subject the agent to adversarial training against potent attacks. In this work, we propose a novel method referred to as Transformed Input-robust RL (TIRL), which explores another avenue to mitigate the impact of adversaries by employing input transformation-based defenses. Specifically, we introduce two principles for applying transformation-based defenses in learning robust RL agents: (1) autoencoder-styled denoising to reconstruct the original state and (2) bounded transformations (bit-depth reduction and vector quantization (VQ)) to achieve close transformed inputs. The transformations are applied to the state before feeding it into the policy network. Extensive experiments on multiple MuJoCo environments demonstrate that input transformation-based defenses, i.e., VQ, defend against several adversaries in the state observations. The official code is available at https://github.com/tunglm2203/tirl
This paper establishes a new and comprehensive theoretical analysis for the application of reinforcement learning (RL) in high-frequency market making. We bridge the modern RL theory and the continuous-time statistical models in high-frequency financial economics. Different with most existing literature on methodological research about developing various RL methods for market making problem, our work is a pilot to provide the theoretical analysis. We target the effects of sampling frequency, and find an interesting tradeoff between error and complexity of RL algorithm when tweaking the values of the time increment $\Delta$ $-$ as $\Delta$ becomes smaller, the error will be smaller but the complexity will be larger. We also study the two-player case under the general-sum game framework and establish the convergence of Nash equilibrium to the continuous-time game equilibrium as $\Delta\rightarrow0$. The Nash Q-learning algorithm, which is an online multi-agent RL method, is applied to solve the equilibrium. Our theories are not only useful for practitioners to choose the sampling frequency, but also very general and applicable to other high-frequency financial decision making problems, e.g., optimal executions, as long as the time-discretization of a continuous-time markov decision process is adopted. Monte Carlo simulation evidence support all of our theories.
Distributional reinforcement learning improves performance by capturing environmental stochasticity, but a comprehensive theoretical understanding of its effectiveness remains elusive. In addition, the intractable element of the infinite dimensionality of distributions has been overlooked. In this paper, we present a regret analysis of distributional reinforcement learning with general value function approximation in a finite episodic Markov decision process setting. We first introduce a key notion of $\textit{Bellman unbiasedness}$ which is essential for exactly learnable and provably efficient distributional updates in an online manner. Among all types of statistical functionals for representing infinite-dimensional return distributions, our theoretical results demonstrate that only moment functionals can exactly capture the statistical information. Secondly, we propose a provably efficient algorithm, $\texttt{SF-LSVI}$, that achieves a tight regret bound of $\tilde{O}(d_E H^{\frac{3}{2}}\sqrt{K})$ where $H$ is the horizon, $K$ is the number of episodes, and $d_E$ is the eluder dimension of a function class.
General intelligence requires quick adaption across tasks. While existing reinforcement learning (RL) methods have made progress in generalization, they typically assume only distribution changes between source and target domains. In this paper, we explore a wider range of scenarios where not only the distribution but also the environment spaces may change. For example, in the CoinRun environment, we train agents from easy levels and generalize them to difficulty levels where there could be new enemies that have never occurred before. To address this challenging setting, we introduce a causality-guided self-adaptive representation-based approach, called CSR, that equips the agent to generalize effectively across tasks with evolving dynamics. Specifically, we employ causal representation learning to characterize the latent causal variables within the RL system. Such compact causal representations uncover the structural relationships among variables, enabling the agent to autonomously determine whether changes in the environment stem from distribution shifts or variations in space, and to precisely locate these changes. We then devise a three-step strategy to fine-tune the causal model under different scenarios accordingly. Empirical experiments show that CSR efficiently adapts to the target domains with only a few samples and outperforms state-of-the-art baselines on a wide range of scenarios, including our simulated environments, CartPole, CoinRun and Atari games.
This paper introduces a novel reinforcement learning (RL) strategy designed to facilitate rapid autonomy transfer by utilizing pre-trained critic value functions from multiple environments. Unlike traditional methods that require extensive retraining or fine-tuning, our approach integrates existing knowledge, enabling an RL agent to adapt swiftly to new settings without requiring extensive computational resources. Our contributions include development of the Multi-Critic Actor-Critic (MCAC) algorithm, establishing its convergence, and empirical evidence demonstrating its efficacy. Our experimental results show that MCAC significantly outperforms the baseline actor-critic algorithm, achieving up to 22.76x faster autonomy transfer and higher reward accumulation. This advancement underscores the potential of leveraging accumulated knowledge for efficient adaptation in RL applications.
Offline reinforcement learning often requires a quality dataset that we can train a policy on. However, in many situations, it is not possible to get such a dataset, nor is it easy to train a policy to perform well in the actual environment given the offline data. We propose using data distillation to train and distill a better dataset which can then be used for training a better policy model. We show that our method is able to synthesize a dataset where a model trained on it achieves similar performance to a model trained on the full dataset or a model trained using percentile behavioral cloning. Our project site is available at $\href{https://datasetdistillation4rl.github.io}{\text{here}}$. We also provide our implementation at $\href{https://github.com/ggflow123/DDRL}{\text{this GitHub repository}}$.
The field of reinforcement learning offers a large variety of concepts and methods to tackle sequential decision-making problems. This variety has become so large that choosing an algorithm for a task at hand can be challenging. In this work, we streamline the process of choosing reinforcement-learning algorithms and action-distribution families. We provide a structured overview of existing methods and their properties, as well as guidelines for when to choose which methods. An interactive version of these guidelines is available online at https://rl-picker.github.io/.
This paper introduces ARCLE, an environment designed to facilitate reinforcement learning research on the Abstraction and Reasoning Corpus (ARC). Addressing this inductive reasoning benchmark with reinforcement learning presents these challenges: a vast action space, a hard-to-reach goal, and a variety of tasks. We demonstrate that an agent with proximal policy optimization can learn individual tasks through ARCLE. The adoption of non-factorial policies and auxiliary losses led to performance enhancements, effectively mitigating issues associated with action spaces and goal attainment. Based on these insights, we propose several research directions and motivations for using ARCLE, including MAML, GFlowNets, and World Models.
In recent years, Multi-Agent Reinforcement Learning (MARL) has found application in numerous areas of science and industry, such as autonomous driving, telecommunications, and global health. Nevertheless, MARL suffers from, for instance, an exponential growth of dimensions. Inherent properties of quantum mechanics help to overcome these limitations, e.g., by significantly reducing the number of trainable parameters. Previous studies have developed an approach that uses gradient-free quantum Reinforcement Learning and evolutionary optimization for variational quantum circuits (VQCs) to reduce the trainable parameters and avoid barren plateaus as well as vanishing gradients. This leads to a significantly better performance of VQCs compared to classical neural networks with a similar number of trainable parameters and a reduction in the number of parameters by more than 97 \% compared to similarly good neural networks. We extend an approach of K\"olle et al. by proposing a Gate-Based, a Layer-Based, and a Prototype-Based concept to mutate and recombine VQCs. Our results show the best performance for mutation-only strategies and the Gate-Based approach. In particular, we observe a significantly better score, higher total and own collected coins, as well as a superior own coin rate for the best agent when evaluated in the Coin Game environment.
As Deep Reinforcement Learning (Deep RL) research moves towards solving large-scale worlds, efficient environment simulations become crucial for rapid experimentation. However, most existing environments struggle to scale to high throughput, setting back meaningful progress. Interactions are typically computed on the CPU, limiting training speed and throughput, due to slower computation and communication overhead when distributing the task across multiple machines. Ultimately, Deep RL training is CPU-bound, and developing batched, fast, and scalable environments has become a frontier for progress. Among the most used Reinforcement Learning (RL) environments, MiniGrid is at the foundation of several studies on exploration, curriculum learning, representation learning, diversity, meta-learning, credit assignment, and language-conditioned RL, and still suffers from the limitations described above. In this work, we introduce NAVIX, a re-implementation of MiniGrid in JAX. NAVIX achieves over 200 000x speed improvements in batch mode, supporting up to 2048 agents in parallel on a single Nvidia A100 80 GB. This reduces experiment times from one week to 15 minutes, promoting faster design iterations and more scalable RL model development.
Offline reinforcement learning has shown promise for solving tasks in safety-critical settings, such as clinical decision support. Its application, however, has been limited by the lack of interpretability and interactivity for clinicians. To address these challenges, we propose the medical decision transformer (MeDT), a novel and versatile framework based on the goal-conditioned reinforcement learning paradigm for sepsis treatment recommendation. MeDT uses the decision transformer architecture to learn a policy for drug dosage recommendation. During offline training, MeDT utilizes collected treatment trajectories to predict administered treatments for each time step, incorporating known treatment outcomes, target acuity scores, past treatment decisions, and current and past medical states. This analysis enables MeDT to capture complex dependencies among a patient's medical history, treatment decisions, outcomes, and short-term effects on stability. Our proposed conditioning uses acuity scores to address sparse reward issues and to facilitate clinician-model interactions, enhancing decision-making. Following training, MeDT can generate tailored treatment recommendations by conditioning on the desired positive outcome (survival) and user-specified short-term stability improvements. We carry out rigorous experiments on data from the MIMIC-III dataset and use off-policy evaluation to demonstrate that MeDT recommends interventions that outperform or are competitive with existing offline reinforcement learning methods while enabling a more interpretable, personalized and clinician-directed approach.
Reinforcement learning practitioners often avoid hierarchical policies, especially in image-based observation spaces. Typically, the single-task performance improvement over flat-policy counterparts does not justify the additional complexity associated with implementing a hierarchy. However, by introducing multiple decision-making levels, hierarchical policies can compose lower-level policies to more effectively generalize between tasks, highlighting the need for multi-task evaluations. We analyze the benefits of hierarchy through simulated multi-task robotic control experiments from pixels. Our results show that hierarchical policies trained with task conditioning can (1) increase performance on training tasks, (2) lead to improved reward and state-space generalizations in similar tasks, and (3) decrease the complexity of fine tuning required to solve novel tasks. Thus, we believe that hierarchical policies should be considered when building reinforcement learning architectures capable of generalizing between tasks.
The rapid advancement of quantum computing (QC) and machine learning (ML) has given rise to the burgeoning field of quantum machine learning (QML), aiming to capitalize on the strengths of quantum computing to propel ML forward. Despite its promise, crafting effective QML models necessitates profound expertise to strike a delicate balance between model intricacy and feasibility on Noisy Intermediate-Scale Quantum (NISQ) devices. While complex models offer robust representation capabilities, their extensive circuit depth may impede seamless execution on extant noisy quantum platforms. In this paper, we address this quandary of QML model design by employing deep reinforcement learning to explore proficient QML model architectures tailored for designated supervised learning tasks. Specifically, our methodology involves training an RL agent to devise policies that facilitate the discovery of QML models without predetermined ansatz. Furthermore, we integrate an adaptive mechanism to dynamically adjust the learning objectives, fostering continuous improvement in the agent's learning process. Through extensive numerical simulations, we illustrate the efficacy of our approach within the realm of classification tasks. Our proposed method successfully identifies VQC architectures capable of achieving high classification accuracy while minimizing gate depth. This pioneering approach not only advances the study of AI-driven quantum circuit design but also holds significant promise for enhancing performance in the NISQ era.
One important property of DIstribution Correction Estimation (DICE) methods is that the solution is the optimal stationary distribution ratio between the optimized and data collection policy. In this work, we show that DICE-based methods can be viewed as a transformation from the behavior distribution to the optimal policy distribution. Based on this, we propose a novel approach, Diffusion-DICE, that directly performs this transformation using diffusion models. We find that the optimal policy's score function can be decomposed into two terms: the behavior policy's score function and the gradient of a guidance term which depends on the optimal distribution ratio. The first term can be obtained from a diffusion model trained on the dataset and we propose an in-sample learning objective to learn the second term. Due to the multi-modality contained in the optimal policy distribution, the transformation in Diffusion-DICE may guide towards those local-optimal modes. We thus generate a few candidate actions and carefully select from them to approach global-optimum. Different from all other diffusion-based offline RL methods, the guide-then-select paradigm in Diffusion-DICE only uses in-sample actions for training and brings minimal error exploitation in the value function. We use a didatic toycase example to show how previous diffusion-based methods fail to generate optimal actions due to leveraging these errors and how Diffusion-DICE successfully avoids that. We then conduct extensive experiments on benchmark datasets to show the strong performance of Diffusion-DICE. Project page at https://ryanxhr.github.io/Diffusion-DICE/.
The deployment of artificial intelligence (AI) in decision-making applications requires ensuring an appropriate level of safety and reliability, particularly in changing environments that contain a large number of unknown observations. To address this challenge, we propose a novel safe reinforcement learning (RL) approach that utilizes an anomalous state sequence to enhance RL safety. Our proposed solution Safe Reinforcement Learning with Anomalous State Sequences (AnoSeqs) consists of two stages. First, we train an agent in a non-safety-critical offline 'source' environment to collect safe state sequences. Next, we use these safe sequences to build an anomaly detection model that can detect potentially unsafe state sequences in a 'target' safety-critical environment where failures can have high costs. The estimated risk from the anomaly detection model is utilized to train a risk-averse RL policy in the target environment; this involves adjusting the reward function to penalize the agent for visiting anomalous states deemed unsafe by our anomaly model. In experiments on multiple safety-critical benchmarking environments including self-driving cars, our solution approach successfully learns safer policies and proves that sequential anomaly detection can provide an effective supervisory signal for training safety-aware RL agents
Interpretability of deep reinforcement learning systems could assist operators with understanding how they interact with their environment. Vector quantization methods -- also called codebook methods -- discretize a neural network's latent space that is often suggested to yield emergent interpretability. We investigate whether vector quantization in fact provides interpretability in model-based reinforcement learning. Our experiments, conducted in the reinforcement learning environment Crafter, show that the codes of vector quantization models are inconsistent, have no guarantee of uniqueness, and have a limited impact on concept disentanglement, all of which are necessary traits for interpretability. We share insights on why vector quantization may be fundamentally insufficient for model interpretability.