analiza aiqbench
This research introduces PAMD, a method that replaces fixed global metrics with adaptive pairwise distances to improve visual reinforcement learning representations. By parameterizing positive-definite metrics conditioned on state pairs, the approach resolves limitations of hand-designed norms and prevents degenerate solutions in latent space modeling. The study demonstrates substantial performance gains across continuous-control tasks when integrated into existing bisimulation-based algorithms.
PAMD: Structured Adaptive Distances for Bisimulation Representations in Visual Reinforcement Learning
arXiv:2607.18004v1 Announce Type: new Abstract: Many visual reinforcement learning (RL) algorithms learn representations by matching latent distances to a behavioral distance induced by reward and transition similarity. In practice, the choice of the latent distance can strongly affect performance: using a fixed, pre-specified global norms (e.g., $\ell_p$ norms or other hand-designed metrics) may be overly restrictive to capture the behavioral distance. In contrast, unconstrained pairwise distances may admit degenerate solutions that drive the metric loss down without improving the representation. To address this gap, we introduce **PAMD: Pairwise Adaptive Mahalanobis Distance**, which parameterizes a positive-definite, pair-conditioned metric for measuring latent state similarity. PAMD is a simple plug-in for existing bisimulation-based methods, offering a more expressive yet structured alternative to fixed, pre-specified latent distances. We empirically validate our method on visual MuJoCo continuous-control tasks, where final performance of several recent bisimulation-based RL algorithms is substantially improved when equipped with the distance we propose.
Abstract:Many visual reinforcement learning (RL) algorithms learn representations by matching latent distances to a behavioral distance induced by reward and transition similarity. In practice, the choice of the latent distance can strongly affect performance: using a fixed, pre-specified global norms (e.g., $\ell_p$ norms or other hand-designed metrics) may be overly restrictive to capture the behavioral distance. In contrast, unconstrained pairwise distances may admit degenerate solutions that drive the metric loss down without improving the representation. To address this gap, we introduce **PAMD: Pairwise Adaptive Mahalanobis Distance**, which parameterizes a positive-definite, pair-conditioned metric for measuring latent state similarity. PAMD is a simple plug-in for existing bisimulation-based methods, offering a more expressive yet structured alternative to fixed, pre-specified latent distances. We empirically validate our method on visual MuJoCo continuous-control tasks, where final performance of several recent bisimulation-based RL algorithms is substantially improved when equipped with the distance we propose.
9 pages, 6 figures, plus appendix. Accepted to the 43rd International Conference on Machine Learning (ICML 2026)
Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2607.18004 [cs.AI]
(or arXiv:2607.18004v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2607.18004 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 261 cuvinte
Subiecte conexe
Developing Healthcare Robotics with GPU-Native Medical Physics Simulation
NVIDIA · 28 iul. 2026
News
Quantum Computing Report · 25 iul. 2026
Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
Microsoft Research · 11 aug. 2026
Amazon is investing in the Lean Focused Research Organization
Amazon Science · 26 iul. 2026