Aug 2026· 2026 IEEE International Conference on Mechatronics and Automation (ICMA)· pp. 510-517· 0 citations· 32 references
Abstract
In Model-Based Reinforcement Learning, Deep Neural Networks are trained to be world models to simulate interactions between agents and their environments. Model biases are inevitable in such world models due to their limited size and the stochastic nature of training datasets. Instead of striving for a perfect world model, we set out to mitigate the impacts of potentially erroneous synthetic data by weighting it based on its uncertainty. To address the low-variance domination issue during weighting, we introduce a practical, easy-to-tune Immersive Weighting that assigns uniform weights to low-variance transitions. Empirical evaluations demonstrate that the proposed method effectively enhances a Dyna-Q framework in our real-world and four of nine simulation tasks, laying the groundwork for future advancements in task-solving efficiency.
ESRL is introduced, an architecture-aware framework that explicitly explores the expert-routing space of MoE models, and preserves high-confidence experts as anchors, and restricts stochastic routing to a plausible candidate pool, thereby retaining reliable computation paths.
Hong-Yi He, Zheng-Wen Lin, Xiao Liu et al.· 0 citations
Model-based reinforcement learning (MBRL) can exploit simulated experience, but a simulator-to-target shift creates a model-selection problem: correcting the simulator and fitting the target directly can each fail under limited target data. We introduce the Model-Corrected World Model (MC-WM), which separates initial t...
CLAW (Context-conditioned Low-rank Adaptation of World models), which addresses the problem of adapting a world model to an unknown test-time environment, drawn from a known environment family, using only a few episodes of interaction by using a hypernetwork to generate low-rank adapters at test time.
Fernando Palafox, David Fridovich-Keil· 0 citations
F orecasting in Non-stationary O ffline RL (F ORL), a framework that combines zero-shot forecasting with the agent’s experience, aims to bridge the gap between offline RL and the complexities of real-world, non-stationary environments.
Suzan Ece, Georg Martius, Emre Ugur et al.· 0 citations
ELBO-based reinforcement learning offers a sampler-agnostic approach to fine-tuning flow matching models with reward feedback. Timestep weighting in ELBO-based RL has large impact on performance, and it also provides a unified view (as we show in this work) to understand prediction losses heuristically chosen in prior...
Qin-Wei Ma, Jing-Zhe Shi, Si-Min Fan et al.· 0 citations
A new robust variant of MCTS that mitigates dynamical model ambiguities to bridge the gap between simulation-based planning and real-world deployment and empirical evidence is provided that this method achieves robust performance in planning problems even under significant ambiguity in the underlying reward distributio...
T. Dam, Kishan Panaganti, Brahim Driss et al.· International Conference on...· 5 citations· ⚡1
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
