Prism: Policy Reuse via Interpretable Strategy Mapping in Reinforcement Learning

• URL: http://arxiv.org/abs/2604.02353v1
• Date: Wed, 04 Mar 2026 22:51:22 GMT
• Title: Prism: Policy Reuse via Interpretable Strategy Mapping in Reinforcement Learning
• Abstract summary: PRISM is a framework that grounds reinforcement learning agents' decisions in discrete, causally validated concepts.<n> PRISM clusters each agent's encoder features into $K$ concepts via K-means.<n>Because concepts causally encode strategy, aligning them via optimal bipartite matching transfers strategic knowledge zero-shot.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We present PRISM (Policy Reuse via Interpretable Strategy Mapping), a framework that grounds reinforcement learning agents' decisions in discrete, causally validated concepts and uses those concepts as a zero-shot transfer interface between agents trained with different algorithms. PRISM clusters each agent's encoder features into $K$ concepts via K-means. Causal intervention establishes that these concepts directly drive - not merely correlate with - agent behavior: overriding concept assignments changes the selected action in 69.4% of interventions ($p = 8.6 \times 10^{-86}$, 2500 interventions). Concept importance and usage frequency are dissociated: the most-used concept (C47, 33.0% frequency) causes only a 9.4% win-rate drop when ablated, while ablating C16 (15.4% frequency) collapses win rate from 100% to 51.8%. Because concepts causally encode strategy, aligning them via optimal bipartite matching transfers strategic knowledge zero-shot. On Go~7$\times$7 with three independently trained agents, concept transfer achieves 69.5%$\pm$3.2% and 76.4%$\pm$3.4% win rate against a standard engine across the two successful transfer pairs (10 seeds), compared to 3.5% for a random agent and 9.2% without alignment. Transfer succeeds when the source policy is strong; geometric alignment quality predicts nothing ($R^2 \approx 0$). The framework is scoped to domains where strategic state is naturally discrete: the identical pipeline on Atari Breakout yields bottleneck policies at random-agent performance, confirming that the Go results reflect a structural property of the domain.

Related papers

• EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents [74.54929751174289]
  EvoSafeHarness is a safety-specific optimization framework that synthesizes a deployable harness for a frozen model in a target domain.<n>It reduces average attack success rate from 45.6% to 10.0% at a 3.3-point utility cost.<n>It also achieves the best score on Agent-SafetyBench for every victim.
  arXiv  Detail & Related papers  (2026-09-05T05:56:41Z)
• Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning [55.264863369127774]
  Current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.<n>We propose Hidden-Align, an auxiliary loss function that aligns the last-layer hidden states of correct rollouts at the anchor token during RL training, with zero overhead in both training and inference.<n>On eight mathematical reasoning benchmarks, Hidden-Align improves average pass@1 over the DAPO baseline by 3.8, 6.2, and 5.4 percentage points on Qwen3-1.7B, 4B, and 14B respectively, with consistent pass@k gains across all three scales.
  arXiv  Detail & Related papers  (2026-06-02T06:51:15Z)
• Robust Multi-Agent Path Finding under Observation Attacks: A Principled Adversarial-Plus-Smoothing Training Recipe [0.0]
  We present two training recipes that keep the same network and the same deployment loop, yet make the policy hold up under perturbed observations.<n>The first recipe, Adv-PPO, trains the shared policy against worst-case perturbations of its own input and selects the checkpoint by performance under adversarial perturbation.<n>The second recipe, Adv-PPO+MACER, fine-tunes that checkpoint with a small on-policy smoothness term whose gradient follows the certified radius of randomized smoothing.
  arXiv  Detail & Related papers  (2026-05-12T03:33:34Z)
• Preserving Disagreement: Architectural Heterogeneity and Coherence Validation in Multi-Agent Policy Simulation [0.0]
  Multi-agent deliberation systems using large language models (LLMs) are increasingly proposed for policy simulation.<n>They suffer from artificial consensus: evaluator agents converge on the same option regardless of their assigned value perspectives.<n>We present the AI Council, a three-phase deliberation framework, and conduct 120 deliberations across two policy scenarios to test two interventions.
  arXiv  Detail & Related papers  (2026-04-29T11:47:28Z)
• See Less, Drive Better: Generalizable End-to-End Autonomous Driving via Foundation Models Stochastic Patch Selection [51.59559387222532]
  Recent advances in end-to-end autonomous driving show that policies trained on patch-aligned features generalize better to Out-of-Distribution (OOD)<n>We present $2.4-Patch-Selection (SPS), a simple yet effective approach for learning policies that are more robust, generalizable, and efficient.
  arXiv  Detail & Related papers  (2026-01-15T18:58:33Z)
• BEDA: Belief Estimation as Probabilistic Constraints for Performing Strategic Dialogue Acts [69.7477004390211]
  We formalize two core acts Adversarial and Alignment, and operationalize them via probabilistic constraints on what an agent may generate.<n>We instantiate this idea in BEDA, a framework that consists of the world set, the belief estimator for belief estimation, and the conditional generator that selects acts and realizes utterances consistent with the inferred beliefs.
  arXiv  Detail & Related papers  (2025-12-31T14:26:55Z)
• Towards a Science of Scaling Agent Systems [79.64446272302287]
  We formalize a definition for agent evaluation and characterize scaling laws as the interplay between agent quantity, coordination structure, modelic, and task properties.<n>We derive a predictive model using coordination metrics, that cross-validated R2=0, enabling prediction on unseen task domains.<n>We identify three effects: (1) a tool-coordination trade-off: under fixed computational budgets, tool-heavy tasks suffer disproportionately from multi-agent overhead, and (2) a capability saturation: coordination yields diminishing or negative returns once single-agent baselines exceed 45%.
  arXiv  Detail & Related papers  (2025-12-09T06:52:21Z)
• Learning When to Switch: Adaptive Policy Selection via Reinforcement Learning [0.0]
  This work demonstrates how an agent can dynamically transition between systematic exploration (coverage) and goal-directed pathfinding (convergence) to improve task performance.<n>Unlike fixed-threshold approaches, the agent uses Q-learning to adapt switching behavior based on coverage percentage and distance to goal.<n>Results show 23-55% improvements in completion time, 83% reduction in runtime variance, and 71% improvement in worst-case scenarios.
  arXiv  Detail & Related papers  (2025-12-06T02:50:32Z)
• Pre-Trained Policy Discriminators are General Reward Models [81.3974586561645]
  We propose a scalable pre-training method named Policy Discriminative Learning (POLAR)<n>POLAR trains a reward model (RM) to discern identical policies and discriminate different ones.<n> Empirical results show that POLAR substantially outperforms traditional non-pre-trained methods.
  arXiv  Detail & Related papers  (2025-07-07T16:56:31Z)
• Provably Efficient Fictitious Play Policy Optimization for Zero-Sum Markov Games with Structured Transitions [145.54544979467872]
  We propose and analyze new fictitious play policy optimization algorithms for zero-sum Markov games with structured but unknown transitions. We prove tight $widetildemathcalO(sqrtK)$ regret bounds after $K$ episodes in a two-agent competitive game scenario. Our algorithms feature a combination of Upper Confidence Bound (UCB)-type optimism and fictitious play under the scope of simultaneous policy optimization.
  arXiv  Detail & Related papers  (2022-07-25T18:29:16Z)

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.