Generalization of synthesizer sound matching to external instrument sounds is highly challenging due to the non-differentiability of sound synthesis process which prohibits the use of out-of-domain sounds for training with synthesis parameter loss. We propose SynthRL, a novel reinforcement learning (RL)-based approach for cross-domain synthesizer sound matching. By incorporating sound similarity into the reward function, SynthRL effectively optimizes synthesis parameters without ground-truth labels, allowing fine-tuning on out-of-domain sounds. Furthermore, we introduce a transformer-based model architecture and reward-based prioritized experience replay to enhance RL training efficiency, considering the unique characteristics of the task. Experimental results demonstrate that SynthRL outperforms state-of-the-art methods on both in-domain and out-of-domain tasks. Further experimental analysis validates the effectiveness of our reward design, showing a strong correlation with human perception of sound similarity.
论文检索
输入标题、作者或关键词，从 1,248 篇学术成果中精准定位
Enhancing Portfolio Optimization via Heuristic-Guided Inverse Reinforcement Learning with Multi-Objective Reward and Graph-based Policy Learning
PDF ↗Portfolio optimization encounters persistent challenges in adapting to dynamic markets due to static assumptions and high-dimensional decision spaces. Although reinforcement learning (RL) has emerged as a potential solution, conventional reward engineering often fails to capture complex market dynamics. Recent advances in deep RL and graph neural networks have attempted to enhance market microstructure modeling. However, these methods still struggle with the systematic integration of financial knowledge. To address the above issues, we propose a novel heuristic-guided inverse reinforcement learning framework for portfolio optimization. Specifically, our framework provides an interpretable expert strategy generation mechanism that takes into account sector diversification and correlation constraints. Then, a multi-objective reward optimization method is adopted to adaptively strike a balance between returns and risks. Furthermore, it also utilizes heterogeneous graph policy learning with hierarchical attention mechanisms to explicitly model inter-stock relationships. Finally, we conduct extensive experiments on real-world financial market data to demonstrate that our framework outperforms several state-of-the-art deep learning and RL baselines in terms of risk-adjusted returns. We provide case studies to showcase the ability of our framework to balance return maximization and risk containment. Our code is publicly available at https://github.com/ChloeWenyiZhang/SmartFolio/.
Efficient exploration of state spaces is critical for the success of deep reinforcement learning (RL). While many methods leverage exploration bonuses to encourage exploration instead of relying solely on extrinsic rewards, these bonus-based approaches often face challenges with learning efficiency and scalability, especially in environments with high-dimensional state spaces. To address these issues, we propose BehavIoral metric-based Latent Exploration (BILE). The core idea is to learn a compact representation within the behavioral metric space that preserves value differences between states. By introducing additional rewards to encourage exploration in this latent space, BILE drives the agent to visit states with higher value diversity and exhibit more behaviorally distinct actions, leading to more effective exploration of the state space. Additionally, we present a novel behavioral metric for efficient and robust training of the state encoder, backed by theoretical guarantees. Extensive experiments on high-dimensional environments, including realistic indoor scenarios in Habitat, robotic tasks in Robosuite, and challenging discrete Minigrid benchmarks, demonstrate the superiority and scalability of our method over other approaches.
We consider the challenging problem of using domain knowledge to improve deep reinforcement learning policies. To this end, we propose LEGIBLE, a novel approach, following a multi-step process, which starts by mining rules from a deep RL policy, constituting a partially symbolic representation. These rules describe which decisions the RL policy makes and which it avoids making. In the second step, we generalize the mined rules using domain knowledge expressed as metamorphic relations. We adapt these relations from software testing to RL to specify expected changes of actions in response to changes in observations. The third step is evaluating generalized rules to determine which generalizations improve performance when enforced. These improvements show weaknesses in the policy, where it has not learned the general rules and thus can be improved by rule guidance. LEGIBLE supported by metamorphic relations provides a principled way of expressing and enforcing domain knowledge about RL environments. We show the efficacy of our approach by demonstrating that it effectively finds weaknesses, accompanied by explanations of these weaknesses, in eleven RL environments and by showcasing that guiding policy execution with rules improves performance w.r.t. gained reward.
InstGAN: Instant Actor-Critic-Driven GAN for De Novo Molecule Generation and Property Optimization
PDF ↗Deep generative models, such as generative adversarial networks (GANs), have been employed for de~novo molecular generation in drug discovery. Most prior studies have utilized reinforcement learning (RL) algorithms, particularly Monte Carlo tree search (MCTS), to handle the discrete nature of molecular representations in GANs. However, due to the inherent instability in training GANs and RL models, along with the high computational cost associated with MCTS sampling, MCTS RL-based GANs struggle to scale to large chemical databases. To tackle these challenges, this study introduces a novel GAN based on actor-critic RL with instant and global rewards, called InstGAN, to generate molecules at the token-level with multi-property optimization. Furthermore, maximized information entropy is leveraged to alleviate the mode collapse. The experimental results demonstrate that InstGAN outperforms other baselines, achieves comparable performance to state-of-the-art models, and efficiently generates molecules with multi-property optimization. The code is available at: https://github.com/tang777777/InstGAN.
Reinforcement Learning (RL) has made significant strides in various domains, and policy gradient methods like Proximal Policy Optimization (PPO) have gained popularity due to their balance in performance, training stability, and computational efficiency. These methods directly optimize policies through gradient-based updates. However, developing effective control policies for environments with complex and non-linear dynamics remains a challenge. High variance in gradient estimates and non-convex optimization landscapes often lead to unstable learning trajectories. Koopman Operator Theory has emerged as a powerful framework for studying non-linear systems through an infinite-dimensional linear operator that acts on a higher-dimensional space of measurement functions. In contrast with their non-linear counterparts, linear systems are simpler, more predictable, and easier to analyze. In this paper, we present Koopman-Inspired Proximal Policy Optimization (KIPPO), which learns an approximately linear latent-space representation of the underlying system’s dynamics while retaining essential features for effective policy learning. This is achieved through a Koopman-approximation auxiliary network that can be added to the baseline policy optimization algorithms without altering the architecture of the core policy or value function. Extensive experimental results demonstrate consistent improvements over the PPO baseline with 6–60% increased performance while reducing variability by up to 91% when evaluated on various continuous control tasks.
Arbitrary style transfer aims to apply the style of any given artistic image to another content image. Still, existing deep learning-based methods often require significant computational costs to generate diverse stylized results. Motivated by this, we propose a novel reinforcement learning-based framework for arbitrary style transfer RLMiniStyler. This framework leverages a unified reinforcement learning policy to iteratively guide the style transfer process by exploring and exploiting stylization feedback, generating smooth sequences of stylized results while achieving model lightweight. Furthermore, we introduce an uncertainty-aware multi-task learning strategy that automatically adjusts loss weights to adapt to the content and style balance requirements at different training stages, thereby accelerating model convergence. Through a series of experiments across image various resolutions, we have validated the advantages of RLMiniStyler over other state-of-the-art methods in generating high-quality, diverse artistic image sequences at a lower cost. Codes are available at https://github.com/fengxiaoming520/RLMiniStyler.
Probabilistic logic shields integrate deep reinforcement learning (RL) with probabilistic logic reasoning to train agents that operate in uncertain environments while giving strong guarantees with respect to logical constraints, such as safety properties. In this demo paper, we introduce a codebase that streamlines the design of custom MiniHack environments where neurosymbolic RL agents leverage probabilistic logic shields to learn safe and interpretable policies with strong guarantees. Our framework allows expert users to easily define and train agents that integrate deep neural policies with probabilistic logic in arbitrarily complex games: from simple exploration to planning and interacting with enemies. Additionally, we provide a web-based platform that showcases our application, offering an interactive interface for the broader community to experiment with and explore the capabilities of neurosymbolic reinforcement learning. This lowers the barrier for researchers and developers, making it accessible for a wider audience to engage with safety-critical RL scenarios.
Towards Robust, Efficient, and Practical Decision-Making: From Reward-Maximizing Deep Reinforcement Learning to Reward-Matching GFlowNets
PDF ↗In this talk, I will present our recent advances in sequential decision-making systems in reward-maximizing deep RL and the emerging reward-matching GFlowNets. The presentation will examine three fundamental challenges: efficiency, robustness, and practical applications.
Reinforcement learning (RL), particularly its combination with deep neural networks referred to as deep RL (DRL), has shown tremendous promise across a wide range of applications, suggesting its potential for enabling the development of sophisticated robotic behaviors. Robotics problems, however, pose fundamental difficulties for the application of RL, stemming from the complexity and cost of interacting with the physical world. These challenges notwithstanding, recent advances have enabled DRL to succeed at some real-world robotic tasks. However, state-of-the-art DRL solutions’ maturity varies significantly across robotic applications. In this talk, I will review the current progress of DRL in real-world robotic applications based on our recent survey paper (with Tang, Abbatematteo, Hu, Chandra, and Martı́n-Martı́n), with a particular focus on evaluating the real-world successes achieved with DRL in realizing several key robotic competencies, including locomotion, navigation, stationary manipulation, mobile manipulation, human-robot interaction, and multi-robot interaction. The analysis aims to identify the key factors underlying those exciting successes, reveal underexplored areas, and provide an overall characterization of the status of DRL in robotics. I will also highlight several important avenues for future work, emphasizing the need for stable and sample-efficient real-world RL paradigms, holistic approaches for discovering and integrating various competencies to tackle complex long-horizon, open-world tasks, and principled development and evaluation procedures. The talk is designed to offer insights for RL practitioners and roboticists toward harnessing RL’s power to create generally capable real-world robotic systems.
Despite significant recent advances in the field of Deep Reinforcement Learning (DRL), such methods typically incur high cost of training to learn effective policies, thus posing cost and safety challenges in many practical applications. To improve the learning efficiency of (D)RL methods, transfer learning (TL) has emerged as a promising approach to leverage prior experience on a source domain to speed learning on a new, but related, target domain. In this paper, we take a novel model-informed approach to TL in DRL by assuming that we have knowledge of both the source and target domain models (which would be the case in the prevalent setting of DRL with simulators). While directly solving either the source or target MDP via solution methods like value iteration is computationally prohibitive, we exploit the fact that if the target and source MDPs differ only due to a small structural change in their rewards, we can apply structured value iteration methods in a procedure we term ModelDiff to solve the much smaller target-source ``Diff'' MDP for a reasonable horizon. This ModelDiff approach can then be integrated into extensions of standard DRL algorithms like ModelDiff (MD) DQN, where it provides enhanced provable lower bound guidance to DQN that often speeds convergence for the positive transfer case while critically avoiding decelerated learning in the negative transfer case. Experiments show that MD-DQN matches or outperforms existing TL methods and baselines in both positive and negative transfer settings.
CTD4 – a Deep Continuous Distributional Actor-Critic Agent with a Kalman Fusion of Multiple Critics
PDF ↗Categorical Distributional Reinforcement Learning (CDRL) has demonstrated superior sample efficiency in learning complex tasks compared to conventional Reinforcement Learning (RL) approaches. However, the practical application of CDRL is encumbered by challenging projection steps, detailed parameter tuning, and domain knowledge. This paper addresses these challenges by introducing a pioneering Continuous Distributional Model-Free RL algorithm tailored for continuous action spaces. The proposed algorithm simplifies the implementation of distributional RL, adopting an actor-critic architecture wherein the critic outputs a continuous probability distribution. Additionally, we propose an ensemble of multiple critics fused through a Kalman fusion mechanism to mitigate overestimation bias. Through a series of experiments, we validate that our proposed method provides a sample-efficient solution for executing complex continuous-control tasks.
As the demands for superior agents grow, the training complexity of Deep Reinforcement Learning (DRL) becomes higher. Thus, accelerating training of DRL has become a major research focus. Dividing the DRL training process into sub-tasks and using parallel computation can effectively reduce training costs. However, current DRL training systems lack sufficient parallelization due to data assignment between sub-task components. This assignment issue has been ignored, but addressing it can further boost training efficiency. Therefore, we propose a high-throughput distributed RL training system called TianJi. It relaxes assignment dependencies between sub-task components and enables event-driven asynchronous communication. Meanwhile, TianJi maintains clear boundaries between sub-task components. To address convergence uncertainty from relaxed assignment dependencies, TianJi proposes a distributed strategy based on the balance of sample production and consumption. The strategy controls the staleness of samples to correct their quality, ensuring convergence. We conducted extensive experiments. TianJi achieves a convergence time acceleration ratio of up to 4.37 compared to related comparison frameworks. When scaled to eight computational nodes, TianJi shows a convergence time speedup of 1.6 and a throughput speedup of 7.13 relative to XingTian, emonstrating its capability to accelerate training and scalability. In data transmission efficiency experiments, TianJi significantly outperforms other frameworks, approaching hardware limits. TianJi also shows effectiveness in on-policy algorithms, achieving convergence time acceleration ratios of 4.36 and 2.95 compared to RLlib and XingTian.
Implicit imitation reinforcement learning (IIRL) is a framework that aims to aid a trainee agent’s learning process via observing the state transitions of a mentor, but without access to the latter's action information. Standard IIRL assumes a shared Markov decision process (MDP) between the mentor and trainee, consequently implying an identical action space. This restriction imposes limitations on the applicability of implicit imitation frameworks in real-life scenarios where, possibly due to variations in physical characteristics, the mentor agent may possess distinct own actions, thereby creating a heterogeneous action setting. In this work, we extend the deep implicit imitation Q-networks (DIIQN) method -an online, model-free, deep RL algorithm for implicit imitation- to allow for heterogeneous action sets between mentor and trainee agents. Equipped with our heterogeneous actions DIIQN (HA-DIIQN) method, a trainee agent can harvest the benefits of IIRL even in heterogeneous action settings, achieving accelerated learning and outperforming non-optimal mentor agents.
In many RL applications, ensuring an agent's actions adhere to constraints is crucial for safety. Most previous methods in Action-Constrained Reinforcement Learning (ACRL) employ a projection layer after the policy network to correct the action. However projection-based methods suffer from issues like the zero gradient problem and higher runtime due to the usage of optimization solvers. Recently methods were proposed to train generative models to learn a differentiable mapping between latent variables and feasible actions to address this issue. However, generative models require training using samples from the constrained action space, which itself is challenging. To address such limitations, first, we define a target distribution for feasible actions based on constraint violation signals, and train normalizing flows by minimizing the KL divergence between an approximated distribution over feasible actions and the target. This eliminates the need to generate feasible action samples, greatly simplifying the flow model learning. Second, we integrate the learned flow model with existing deep RL methods, which restrict it to exploring only the feasible action space. Third, we extend our approach beyond ACRL to handle state-wise constraints by learning the constraint violation signal from the environment. Empirically, our approach has significantly fewer constraint violations while achieving similar or better quality in several control tasks than previous best methods.
In reinforcement learning, especially in sparse-reward domains, many environment steps are required to observe reward information. In order to increase the frequency of such observations, "potential-based reward shaping" (PBRS) has been proposed as a method of providing a more dense reward signal while leaving the optimal policy invariant. However, the required potential function must be carefully designed with task-dependent knowledge to not deter training performance. In this work, we propose a bootstrapped method of reward shaping, termed BS-RS, in which the agent's current estimate of the state-value function acts as the potential function for PBRS. We provide convergence proofs for the tabular setting, give insights into training dynamics for deep RL, and show that the proposed method improves training speed in the Atari suite.
An issue concerning the use of deep reinforcement learning (RL) agents is whether they can be trusted to perform reliably when deployed, as training environments may not reflect real-life environments. Anticipating instances outside their training scope, learning-enabled systems are often equipped with out-of-distribution (OOD) detectors that alert when a trained system encounters a state it does not recognize or in which it exhibits uncertainty. There exists limited work conducted on the problem of OOD detection within RL, with prior studies being unable to achieve a consensus on the definition of OOD execution within the context of RL. By framing our problem using a Markov Decision Process, we assume there is a transition distribution mapping each state-action pair to another state with some probability. Based on this, we consider the following definition of OOD execution within RL: A transition is OOD if its probability during real-life deployment differs from the transition distribution encountered during training. As such, we utilize conditional variational autoencoders (CVAE) to approximate the transition dynamics of the training environment and implement a conformity-based detector using reconstruction loss that is able to guarantee OOD detection with a pre-determined confidence level. We evaluate our detector by adapting existing benchmarks and compare it with existing OOD detection models for RL.
The ability to estimate temporal relationships is critical for both animals and artificial agents. Cognitive science and neuroscience provide remarkable insights into behavioral and neural aspects of temporal credit assignment. In particular, scale invariance of learning dynamics, observed in behavior and supported by neural data, is one of the key principles that governs animal perception: proportional rescaling of temporal relationships does not alter the overall learning efficiency. Here we integrate a computational neuroscience model of scale invariant memory into deep reinforcement learning (RL) agents. We first provide a theoretical analysis and then demonstrate through experiments that such agents can learn robustly across a wide range of temporal scales, unlike agents built with commonly used recurrent memory architectures such as LSTM. This result illustrates that incorporating computational principles from neuroscience and cognitive science into deep neural networks can enhance adaptability to complex temporal dynamics, mirroring some of the core properties of human learning.
Autonomous Goal Detection and Cessation in Reinforcement Learning: A Case Study on Source Term Estimation
PDF ↗Reinforcement Learning has revolutionized decision-making processes in dynamic environments, yet it often struggles with autonomously detecting and achieving goals without clear feedback signals. For example, in a Source Term Estimation problem, the lack of precise environmental information makes it challenging to provide clear feedback signals and to define and evaluate how the source's location is determined. To address this challenge, the Autonomous Goal Detection and Cessation (AGDC) module was developed, enhancing various RL algorithms by incorporating a self-feedback mechanism for autonomous goal detection and cessation upon task completion. Our method effectively identifies and ceases undefined goals by approximating the agent's belief, significantly enhancing the capabilities of RL algorithms in environments with limited feedback. To validate effectiveness of our approach, we integrated AGDC with deep Q-Network, proximal policy optimization, and deep deterministic policy gradient algorithms, and evaluated its performance on the Source Term Estimation problem. The experimental results showed that AGDC-enhanced RL algorithms significantly outperformed traditional statistical methods such as infotaxis, entrotaxis, and dual control for exploitation and exploration, as well as a non-statistical random action selection method. These improvements were evident in terms of success rate, mean traveled distance, and search time, highlighting AGDC's effectiveness and efficiency in complex, real-world scenarios.
Retrieval-augmented generation (RAG) systems have advanced large language models (LLMs) in complex deep search scenarios requiring multi-step reasoning and iterative information retrieval. However, existing approaches face critical limitations that lack high-quality training trajectories or suffer from the distributional mismatches in simulated environments and prohibitive computational costs for real-world deployment. This paper introduces SimpleDeepSearcher, a lightweight yet effective framework that bridges this gap through strategic data engineering rather than complex training paradigms. Our approach synthesizes high-quality training data by simulating realistic user interactions in live web search environments, coupled with a multi-criteria curation strategy that optimizes the diversity and quality of input and output side. Experiments on five benchmarks across diverse domains demonstrate that SFT on only 871 curated samples yields significant improvements over RL-based baselines. Our work establishes SFT as a viable pathway by systematically addressing the data-scarce bottleneck, offering practical insights for efficient deep search systems. Our anonymous code is available at https://github.com/RUCAIBox/SimpleDeepSearcher