Research map / Reinforcement learning / Policy gradient and actor-critic

ODE-based Smoothing Neural Network for Reinforcement Learning Tasks

ICLR 2025 · Reinforcement learning › Policy gradient and actor-critic · approach: Regularized scalable deep value learning

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes SmODE, an ODE-based smoothing neural network that makes RL policy actions smooth by addressing both high-frequency input noise and unconstrained Lipschitz constants.

Problem

Deep RL policies often produce non-smooth control actions due to high-frequency input noise and unconstrained Lipschitz constants in neural networks, which hurts optimal control performance.

Approach

An Smooth ODE (SmODE) network that embeds ordinary differential equation dynamics into the policy so that it simultaneously filters high-frequency input noise and constrains the network's Lipschitz constant, yielding smooth control actions.

Closest papers by meaning

• LipsNet++: Unifying Filter and Controller into a Policy Network (ICML 2025 · similarity 0.89)
• Enhancing Robustness in Deep Reinforcement Learning: A Lyapunov Exponent Approach (NeurIPS 2024 · similarity 0.87)
• The Ladder in Chaos: Improving Policy Learning by Harnessing the Parameter Evolving Path in A Low-dimensional Space (NeurIPS 2024 · similarity 0.85)
• Efficient Reinforcement Learning by Discovering Neural Pathways (NeurIPS 2024 · similarity 0.85)
• Certifying Stability of Reinforcement Learning Policies using Generalized Lyapunov Functions (NeurIPS 2025 · similarity 0.85)
• Proxy Target: Bridging the Gap Between Discrete Spiking Neural Networks and Continuous Control (NeurIPS 2025 · similarity 0.85)