Arithmetic Actor Heads and Training Stabilization for Out-of-Distribution Reinforcement Learning

arXiv:2610.05143 · cs.LG · Submitted 2026-10-04 · Read on arXiv

cs.LG

Submitted: 2026-10-04

Updated: 2026-10-04

Code: https://github.com/erzhu419/CS-BAPR

Terminology

• Actor — 1× in this paper · explained in 3 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes

Sources

• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• BAPR: Bayesian amnesic piecewise-robust reinforcement learning for non-stationary continuous control
• RE-SAC: Disentangling aleatoric and epistemic risks in bus fleet control: A stable and robust ensemble DRL approach
• Neural Arithmetic Logic Units
• Neural Arithmetic Units
• Bayesian Online Changepoint Detection
• Learning Quadrupedal Locomotion over Challenging Terrain
• Natural Actor-Critic for Robust Reinforcement Learning with Function Approximation
• Model-Free Robust $\phi$-Divergence Reinforcement Learning Using Both Offline and Online Data
• Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability
• Learning Equations for Extrapolation and Control
• KAN: Kolmogorov-Arnold Networks
• Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
• SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning
• AI Feynman: a Physics-Inspired Method for Symbolic Regression
• Invariant Risk Minimization
• Towards Causal Representation Learning
• A Survey on Causal Reinforcement Learning
• On the failure of ReLU activation for physics-informed machine learning
• Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks