arXiv Machine Learning September 14, 2026 By Ben Opperman, Eduardo Alonso, Esther Mondrag\'on

Groupoid-Based Internal State Representations for Reinforcement Learning with Local Symmetries

Read the original on arXiv Machine Learning →

The paper introduces a reinforcement learning framework that uses groupoids to model local, state-dependent symmetries, allowing agents to discover equivalence structures during interaction. By maintaining orbit representatives and transporters that map raw states to canonical forms, learning and decision-making occur in a symmetry-reduced space while preserving local distinctions. Experiments show that this approach improves sample efficiency and convergence in dense, large-scale environments with partial symmetries, outperforming standard Q‑learning.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

arXiv AI

Sep 24

Categorical Internalisation of Environmental Groupoids for Generalisable POMDP Solving

The paper proposes using category theory to structure reinforcement learning in high‑dimensional, partially observable environments. By partitioning the state space into equivalence classes (symmetry orbits) and treating each class as a groupoid with a canonical representative, the agent can share learning across similar states, reducing redundancy and improving sample efficiency. Experiments on partially observable benchmarks show that this orbit‑based partitioning consistently enhances performance in environments with latent symmetry.

By Ben Opperman, Eduardo Alonso, Esther Mondrag\'on

agents reinforcement-learning benchmarks

arXiv Machine Learning

4d ago

Sample Complexity of Equivariant Reinforcement Learning

arXiv:2609.36421v1 Announce Type: cross Abstract: Reinforcement learning (RL) is a powerful framework for robotic control, yet its practical application is often hindered by high sample complexity. T...

By Rayan Mazouz, Haibo Zhao, Chris Hillar, Christian Shewmake

reinforcement-learning robotics

arXiv Machine Learning

Jun 5

Reflex: Reinforcement Learning with Reflection Symmetry Exploitation in State-Based Continuous Control

arXiv:2605. 23415v2 Announce Type: replace Abstract: Reinforcement learning has long struggled with poor sample efficiency.

By Shuai Zhen, Yifan Zhang, Yuling Wang, Yanhua Yu

reinforcement-learning benchmarks

arXiv Machine Learning

Jun 2

Task-Induced Representational Invariances Depend on Learning Objective in Deep RL

arXiv:2606. 01868v1 Announce Type: new Abstract: Reinforcement Learning (RL) has long served as a model for goal-directed animal behavior in neuroscience.

By Manu Srinath Halvagal, Sebastian Lee, SueYeon Chung

llms reinforcement-learning fine-tuning

Hugging Face Trending Papers

Jul 13

SKooP: Symmetric Koopman Predictions for Faster and More Generalizable Legged Robot Locomotion with Reinforcement Learning

Reinforcement learning (RL) algorithms classically suffer from poor sample efficiency. In robotics, a recent line of work has emerged addressing this problem by encoding physics priors in the learning process.

agents reinforcement-learning robotics benchmarks

arXiv Machine Learning

Jun 17

Performance-Driven Environment Abstraction with Multi-Timescale Learning

arXiv:2606. 17377v1 Announce Type: new Abstract: We study performance-driven environment abstraction for decision-making in large Markov decision processes.

By Yue Guan, Dipankar Maity, Panagiotis Tsiotras

reinforcement-learning

One email a morning, machine-written

Related stories

Categorical Internalisation of Environmental Groupoids for Generalisable POMDP Solving

Sample Complexity of Equivariant Reinforcement Learning

Reflex: Reinforcement Learning with Reflection Symmetry Exploitation in State-Based Continuous Control

Task-Induced Representational Invariances Depend on Learning Objective in Deep RL

SKooP: Symmetric Koopman Predictions for Faster and More Generalizable Legged Robot Locomotion with Reinforcement Learning

Performance-Driven Environment Abstraction with Multi-Timescale Learning