Double Horizon Model-Based Policy Optimization

• URL: http://arxiv.org/abs/2512.15439v1
• Date: Wed, 17 Dec 2025 13:37:23 GMT
• Title: Double Horizon Model-Based Policy Optimization
• Authors: Akihiro Kubo, Paavo Parmas, Shin Ishii,
• Abstract summary: Long "distribution rollout" (DR) generates on-policy state samples for mitigating distribution shift.<n>Short "training rollout" (TR) offers accurate value gradient estimation with stable gradient updates.<n>We demonstrate that the double-horizon approach effectively balances distribution shift, model bias, and gradient instability.
• Score: 3.257742896258293
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Model-based reinforcement learning (MBRL) reduces the cost of real-environment sampling by generating synthetic trajectories (called rollouts) from a learned dynamics model. However, choosing the length of the rollouts poses two dilemmas: (1) Longer rollouts better preserve on-policy training but amplify model bias, indicating the need for an intermediate horizon to mitigate distribution shift (i.e., the gap between on-policy and past off-policy samples). (2) Moreover, a longer model rollout may reduce value estimation bias but raise the variance of policy gradients due to backpropagation through multiple steps, implying another intermediate horizon for stable gradient estimates. However, these two optimal horizons may differ. To resolve this conflict, we propose Double Horizon Model-Based Policy Optimization (DHMBPO), which divides the rollout procedure into a long "distribution rollout" (DR) and a short "training rollout" (TR). The DR generates on-policy state samples for mitigating distribution shift. In contrast, the short TR leverages differentiable transitions to offer accurate value gradient estimation with stable gradient updates, thereby requiring fewer updates and reducing overall runtime. We demonstrate that the double-horizon approach effectively balances distribution shift, model bias, and gradient instability, and surpasses existing MBRL methods on continuous-control benchmarks in terms of both sample efficiency and runtime.

Related papers

• SENTINEL: Stagewise Integrity Verification for Pipeline Parallel Decentralized Training [54.8494905524997]
  Decentralized training introduces critical security risks when executed across untrusted, geographically distributed nodes.<n>We propose SENTINEL, a verification mechanism for pipeline parallelism (PP) training without duplication.<n>Experiments demonstrate successful training of up to 4B- parameter LLMs across untrusted distributed environments with up to 176 workers while maintaining model convergence and performance.
  arXiv  Detail & Related papers  (2026-03-03T23:51:10Z)
• Less is More: Clustered Cross-Covariance Control for Offline RL [13.198112768636207]
  A fundamental challenge in offline reinforcement learning is distributional shift.<n>We propose partitioned buffer sampling that restricts updates to localized replay partitions.<n>We also introduce an explicit gradient-based corrective penalty that cancels the covariance induced bias within each update.
  arXiv  Detail & Related papers  (2026-01-28T16:55:04Z)
• Parallel Diffusion Solver via Residual Dirichlet Policy Optimization [88.7827307535107]
  Diffusion models (DMs) have achieved state-of-the-art generative performance but suffer from high sampling latency due to their sequential denoising nature.<n>Existing solver-based acceleration methods often face significant image quality degradation under a low-dimensional budget.<n>We propose the Ensemble Parallel Direction solver (dubbed as EPD-EPr), a novel ODE solver that mitigates these errors by incorporating multiple gradient parallel evaluations in each step.
  arXiv  Detail & Related papers  (2025-12-28T05:48:55Z)
• Harnessing Bounded-Support Evolution Strategies for Policy Refinement [3.3656696418661975]
  Triangular-Distribution ES pairs bounded triangular noise with a centered-rank finite-difference estimator to deliver stable, parallelizable, gradient-free updates.<n>In a two-stage pipeline - PPO pretraining followed by TD-ES refinement - this preserves early sample efficiency while enabling robust late-stage gains.<n>Across a suite of robotic manipulation tasks, TD-ES raises success rates by 26.5% relative to PPO and greatly reduces variance, offering a simple, compute-light path to reliable refinement.
  arXiv  Detail & Related papers  (2025-11-13T03:35:52Z)
• BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping [69.74252624161652]
  We propose BAlanced Policy Optimization with Adaptive Clipping (BAPO)<n>BAPO dynamically adjusts clipping bounds to adaptively re-balance positive and negative contributions, preserve entropy, and stabilize RL optimization.<n>On AIME 2024 and AIME 2025 benchmarks, our 7B BAPO model surpasses open-source counterparts such as SkyWork-OR1-7B.
  arXiv  Detail & Related papers  (2025-10-21T12:55:04Z)
• Stabilizing Policy Gradients for Sample-Efficient Reinforcement Learning in LLM Reasoning [77.92320830700797]
  Reinforcement Learning has played a central role in enabling reasoning capabilities of Large Language Models.<n>We propose a tractable computational framework that tracks and leverages curvature information during policy updates.<n>The algorithm, Curvature-Aware Policy Optimization (CAPO), identifies samples that contribute to unstable updates and masks them out.
  arXiv  Detail & Related papers  (2025-10-01T12:29:32Z)
• Half-order Fine-Tuning for Diffusion Model: A Recursive Likelihood Ratio Optimizer [16.103949557802988]
  probabilistic diffusion model (DM) generates content by inferencing through a chain structure.<n>Modern methods are either based on Reinforcement Learning (RL) or truncated Backpropagation (BP)<n>We propose the Recursive Likelihood Ratio (RLR) fine-tuning paradigm for DM.
  arXiv  Detail & Related papers  (2025-02-02T03:00:26Z)
• TransFusion: Covariate-Shift Robust Transfer Learning for High-Dimensional Regression [11.040033344386366]
  We propose a two-step method with a novel fused-regularizer to improve the learning performance on a target task with limited samples. Nonasymptotic bound is provided for the estimation error of the target model. We extend the method to a distributed setting, allowing for a pretraining-finetuning strategy.
  arXiv  Detail & Related papers  (2024-04-01T14:58:16Z)
• Time-series Generation by Contrastive Imitation [87.51882102248395]
  We study a generative framework that seeks to combine the strengths of both: Motivated by a moment-matching objective to mitigate compounding error, we optimize a local (but forward-looking) transition policy. At inference, the learned policy serves as the generator for iterative sampling, and the learned energy serves as a trajectory-level measure for evaluating sample quality.
  arXiv  Detail & Related papers  (2023-11-02T16:45:25Z)
• Model-Based Reparameterization Policy Gradient Methods: Theory and Practical Algorithms [88.74308282658133]
  Reization (RP) Policy Gradient Methods (PGMs) have been widely adopted for continuous control tasks in robotics and computer graphics. Recent studies have revealed that, when applied to long-term reinforcement learning problems, model-based RP PGMs may experience chaotic and non-smooth optimization landscapes. We propose a spectral normalization method to mitigate the exploding variance issue caused by long model unrolls.
  arXiv  Detail & Related papers  (2023-10-30T18:43:21Z)
• Distributionally Robust Models with Parametric Likelihood Ratios [123.05074253513935]
  Three simple ideas allow us to train models with DRO using a broader class of parametric likelihood ratios. We find that models trained with the resulting parametric adversaries are consistently more robust to subpopulation shifts when compared to other DRO approaches.
  arXiv  Detail & Related papers  (2022-04-13T12:43:12Z)
• MDPGT: Momentum-based Decentralized Policy Gradient Tracking [29.22173174168708]
  We propose a momentum-based decentralized policy gradient tracking (MDPGT) for multi-agent reinforcement learning. MDPGT achieves the best available sample complexity of $mathcalO(N-1epsilon-3)$ for converging to an $epsilon-stationary point of the global average of $N$ local performance functions. This outperforms the state-of-the-art sample complexity in decentralized model-free reinforcement learning.
  arXiv  Detail & Related papers  (2021-12-06T06:55:51Z)
• Comparing Probability Distributions with Conditional Transport [63.11403041984197]
  We propose conditional transport (CT) as a new divergence and approximate it with the amortized CT (ACT) cost. ACT amortizes the computation of its conditional transport plans and comes with unbiased sample gradients that are straightforward to compute. On a wide variety of benchmark datasets generative modeling, substituting the default statistical distance of an existing generative adversarial network with ACT is shown to consistently improve the performance.
  arXiv  Detail & Related papers  (2020-12-28T05:14:22Z)

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.