paper-with-me

Papers

CEM-GD: Cross-Entropy Method with Gradient Descent Planner for Model-Based Reinforcement Learning

2021-12-14 · Kevin Huang, Sahin Lale, Ugo Rosolia, Yuanyuan Shi, Anima Anandkumar

Current state-of-the-art model-based reinforcement learning algorithms use trajectory sampling methods, such as the Cross-Entropy Method (CEM), for planning in continuous control settings. These zeroth-order optimizers require sampling a large number of trajectory rollouts to select an optimal action, which scales poorly for large prediction horizons or high dimensional action spaces. First-order methods that use the gradients of the rewards with respect to the actions as an update can mitigate this issue, but suffer from local optima due to the non-convex optimization landscape. To overcome these issues and achieve the best of both worlds, we propose a novel planner, Cross-Entropy Method with Gradient Descent (CEM-GD), that combines first-order methods with CEM. At the beginning of execution, CEM-GD uses CEM to sample a significant amount of trajectory rollouts to explore the optimization landscape and avoid poor local minima. It then uses the top trajectories as initialization for gradient descent and applies gradient updates to each of these trajectories to find the optimal action sequence. At each subsequent time step, however, CEM-GD samples much fewer trajectories from CEM before applying gradient updates. We show that as the dimensionality of the planning problem increases, CEM-GD maintains desirable performance with a constant small number of samples by using the gradient information, while avoiding local optima using initially well-sampled trajectories. Furthermore, CEM-GD achieves better performance than CEM on a variety of continuous control benchmarks in MuJoCo with 100x fewer samples per time step, resulting in around 25% less computation time and 10% less memory usage. The implementation of CEM-GD is available at $\href{https://github.com/KevinHuang8/CEM-GD}{\text{https://github.com/KevinHuang8/CEM-GD}}$.

📄 PDF Abstract BibTeX arXiv:2112.07746

Code (1)

kevinhuang8/cem-gd 공식 구현 pytorch

Tasks

continuous-controlContinuous ControlModel-based Reinforcement LearningMuJoCoReinforcement Learning (RL)

Similar Papers 제목 키워드 기반

A proof of convergence of inverse reinforcement learning for multi-objective optimization

2023-05-10 · Akira Kitaoka, Riki Eto

We show the convergence of Wasserstein inverse reinforcement learning for multi-objective optimizations with the projective subgradient method by formulating an inverse problem of the multi-objective optimization problem…

reinforcement-learningReinforcement Learning

Low Entropy Communication in Multi-Agent Reinforcement Learning

2023-02-10 · Lebin Yu, Yunbo Qiu, Qiexiang Wang, Xudong Zhang 외

Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thu…

Multi-agent Reinforcement Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Model-Predictive Control via Cross-Entropy and Gradient-Based Optimization

2020-04-19 · L4DC 2020 6 · Homanga Bharadhwaj, Kevin Xie, Florian Shkurti

Recent works in high-dimensional model-predictive control and model-based reinforcement learning with learned dynamics and reward models have resorted to population-based optimization methods, such as the Cross-Entropy M…

Model-based Reinforcement LearningModel Predictive ControlReinforcement Learning

Mean-Field Microcanonical Gradient Descent

2024-03-13 · Marcus Häggbom, Morten Karlsmark, Joakim andén

Microcanonical gradient descent is a sampling procedure for energy-based models allowing for efficient sampling of distributions in high dimension. It works by transporting samples from a high-entropy distribution, such …

Time Series

Flow Matching Policy Optimization with Mirror Descent and Entropy Constraints

2026-03-18 · Ting Gao, Stavros Orfanoudakis, Nan Lin, Winnie Daamen 외 arxiv

Balancing policy expressiveness with the exploration-exploitation trade-off is a core challenge in online Reinforcement Learning (RL). While Stochastic Differential Equation (SDE)-based diffusion policies can represent c…

Reinforcement Learning