Belief-State RWKV for Reinforcement Learning under Partial Observability

• URL: http://arxiv.org/abs/2604.09671v1
• Date: Wed, 01 Apr 2026 22:28:38 GMT
• Title: Belief-State RWKV for Reinforcement Learning under Partial Observability
• Abstract summary: We propose a stronger formulation of RL on top of RWKV-style recurrent sequence models.<n>We maintain a compact uncertainty-aware state b_t = (_t, _t) derived from RWKV-style recurrent statistics.
• Score: 0.7364191922317778
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We propose a stronger formulation of RL on top of RWKV-style recurrent sequence models, in which the fixed-size recurrent state is explicitly interpreted as a belief state rather than an opaque hidden vector. Instead of conditioning policy and value on a single summary h_t, we maintain a compact uncertainty-aware state b_t = (μ_t, Σ_t) derived from RWKV-style recurrent statistics and let control depend on both memory and uncertainty. This design targets a key weakness of plain fixed-state policies in partially observed settings: they may store evidence, but not necessarily confidence. We present the method, a theoretical program, and a pilot RL experiment with hidden episode-level observation noise together with a test-time noise sweep. The pilot shows that belief-state policies nearly match the best recurrent baseline overall while slightly improving return on the hardest in-distribution regime and under a held-out noise shift. Additional ablations show that this simple belief readout is currently stronger than two more structured extensions, namely gated memory control and privileged belief targets, underscoring the need for richer benchmarks.

Related papers

• ARMOR: Stabilizing On-Policy LLM RL with Off-Policy Anchor Samples [65.74946647574696]
  Reinforcement learning (RL) has significantly enhanced the reasoning capabilities of large language models (LLMs)<n>In this work, we investigate a critical source of this instability: over-optimization.<n>We propose ARMOR, a framework that shifts the paradigm from passive penalty to active sample stabilization.
  arXiv  Detail & Related papers  (2026-07-11T21:22:46Z)
• SAGE: Shaping Anchors for Guided Exploration in RLVR of LLMs [55.46289074417954]
  reinforcement learning with verifiable rewards (RLVR) reliably improves pass@1 on reasoning tasks, yet often fails to yield comparable gains in pass@k.<n>We argue that a central structural constraint arises from reverse-KL regularization, which stabilizes training but inherently anchors the policy to the reference distribution.<n>We propose SAGE, a principled framework that enables controllable empirical support expansion by reshaping the reverse-KL anchor distribution itself.
  arXiv  Detail & Related papers  (2026-05-15T07:42:21Z)
• AdaGamma: State-Dependent Discounting for Temporal Adaptation in Reinforcement Learning [13.721410595245894]
  State-dependent discounting is conceptually appealing, but naive deep actor--critic implementations can degenerate toward TD-error collapse.<n>We propose AdaGamma, a practical deep actor--critic method for state-dependent discounting that learns a state-dependent discount function together with a return-consistency objective.
  arXiv  Detail & Related papers  (2026-05-07T12:42:46Z)
• What If Consensus Lies? Selective-Complementary Reinforcement Learning at Test Time [57.533031432715084]
  Test-Time Reinforcement Learning (TTRL) enables Large Language Models (LLMs) to enhance reasoning capabilities on unlabeled test streams.<n>Existing TTRL methods rely exclusively on positive pseudo-labeling strategies.<n>We propose SCRL, a robust test-time reinforcement learning framework that effectively mitigates label noise amplification.
  arXiv  Detail & Related papers  (2026-03-20T11:47:12Z)
• Fairness Begins with State: Purifying Latent Preferences for Hierarchical Reinforcement Learning in Interactive Recommendation [23.667292152163068]
  We present textbfDSRM-HRL, a framework that reformulates fairness-aware recommendation as a latent state purification problem.<n>We show that DSRM-HRL effectively breaks the "rich-get-richer" feedback loop, achieving a superior frontier between recommendation utility and exposure equity.
  arXiv  Detail & Related papers  (2026-03-04T08:14:21Z)
• Observationally Informed Adaptive Causal Experimental Design [55.998153710215654]
  We propose Active Residual Learning, a new paradigm that leverages the observational model as a foundational prior.<n>This approach shifts the experimental focus from learning target causal quantities from scratch to efficiently estimating the residuals required to correct observational bias.<n> Experiments on synthetic and semi-synthetic benchmarks demonstrate that R-Design significantly outperforms baselines.
  arXiv  Detail & Related papers  (2026-03-04T06:52:37Z)
• VI-CuRL: Stabilizing Verifier-Independent RL Reasoning via Confidence-Guided Variance Reduction [55.04308051033549]
  Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a dominant paradigm for enhancing Large Language Models (LLMs) reasoning.<n>We introduceVerifier-Independent Curriculum Reinforcement Learning (VI-CuRL), a framework that leverages the model's intrinsic confidence to construct a curriculum independent from external verifiers.
  arXiv  Detail & Related papers  (2026-02-13T03:40:52Z)
• Knockoff-Guided Compressive Sensing: A Statistical Machine Learning Framework for Support-Assured Signal Recovery [22.20955211690874]
  This paper introduces a novel Knockoff-guided compressive sensing framework, referred to as TheName.<n>It enhances signal recovery by leveraging precise false discovery rate (FDR) control during the support identification phase.<n>In simulation studies, our method improves F1-score by up to 3.9x over baseline methods, attributed to principled false discovery rate (FDR) control and enhanced support recovery.
  arXiv  Detail & Related papers  (2025-05-30T15:50:58Z)
• ConCISE: Confidence-guided Compression in Step-by-step Efficient Reasoning [64.93140713419561]
  Large Reasoning Models (LRMs) perform strongly in complex reasoning tasks via Chain-of-Thought (CoT) prompting, but often suffer from verbose outputs.<n>Existing fine-tuning-based compression methods either operate post-hoc pruning, risking disruption to reasoning coherence, or rely on sampling-based selection.<n>We introduce ConCISE, a framework designed to generate concise reasoning chains, integrating Confidence Injection to boost reasoning confidence, and Early Stopping to terminate reasoning when confidence is sufficient.
  arXiv  Detail & Related papers  (2025-05-08T01:40:40Z)
• Rethinking State Disentanglement in Causal Reinforcement Learning [78.12976579620165]
  Causality provides rigorous theoretical support for ensuring that the underlying states can be uniquely recovered through identifiability. We revisit this research line and find that incorporating RL-specific context can reduce unnecessary assumptions in previous identifiability analyses for latent states. We propose a novel approach for general partially observable Markov Decision Processes (POMDPs) by replacing the complicated structural constraints in previous methods with two simple constraints for transition and reward preservation.
  arXiv  Detail & Related papers  (2024-08-24T06:49:13Z)
• On Minimizing Adversarial Counterfactual Error in Adversarial RL [18.044879441434432]
  adversarial noise poses significant risks in safety-critical scenarios.<n>We introduce a novel objective called Adversarial Counterfactual Error (ACoE)<n>Our method significantly outperforms current state-of-the-art approaches for addressing adversarial RL challenges.
  arXiv  Detail & Related papers  (2024-06-07T08:14:24Z)
• Exploring the Training Robustness of Distributional Reinforcement Learning against Noisy State Observations [7.776010676090131]
  State observations that an agent observes may contain measurement errors or adversarial noises, misleading the agent to take suboptimal actions or even collapse while training. In this paper, we study the training robustness of distributional Reinforcement Learning (RL), a class of state-of-the-art methods that estimate the whole distribution, as opposed to only the expectation, of the total return.
  arXiv  Detail & Related papers  (2021-09-17T22:37:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.