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

Generalized Policy Iteration using Tensor Approximation for Hybrid Control

ICLR 2024 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees

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 TTPI, a generalized policy iteration method that uses Tensor Train low-rank approximation of state-value and advantage functions for hybrid (discrete-continuous) action control.

Problem

Control of dynamic systems with hybrid (mixed discrete and continuous) actions is challenging in robotics.

Approach

Uses Tensor Train low-rank tensor approximation to represent the state-value and advantage functions within an approximate dynamic programming / generalized policy iteration framework, enabling efficient handling of hybrid actions.

Closest papers by meaning

• Guided Policy Search via Approximate Mirror Descent (NeurIPS 2016 · similarity 0.86)
• Tra-MoE: Learning Trajectory Prediction Model from Multiple Domains for Adaptive Policy Conditioning (CVPR 2025 · similarity 0.85)
• Diffusion Policy Policy Optimization (ICLR 2025 · similarity 0.84)
• Deterministic Policy Gradient Primal-Dual Methods for Continuous-Space Constrained MDPs (AAAI 2025 · similarity 0.84)
• Neural Time-Reversed Generalized Riccati Equation (AAAI 2024 · similarity 0.83)
• Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation (ICLR 2025 · similarity 0.83)