Constrained Exploration

Constrained exploration in machine learning focuses on efficiently searching for optimal solutions within predefined boundaries or limitations, improving the performance of algorithms while avoiding undesirable states or actions. Current research explores this challenge using diverse methods, including modified stochastic gradient descent algorithms, contrastive learning for trajectory optimization in large language model agents, and adaptive trajectory-constrained strategies in reinforcement learning, often incorporating techniques like reward shaping and data augmentation. These advancements are crucial for enhancing the efficiency and reliability of various applications, from robotics and autonomous systems to solving complex optimization problems like routing, where exploring infeasible solutions can be computationally expensive or even impossible.

Papers

May 13, 2024

Constrained Exploration via Reflected Replica Exchange Stochastic Gradient Langevin Dynamics
Haoyang Zheng, Hengrong Du, Qi Feng, Wei Deng, Guang Lin
Multi Modal Non Convex Constrained Exploration

March 4, 2024

Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents
Yifan Song, Da Yin, Xiang Yue, Jie Huang, Sujian Li, Bill Yuchen Lin
LLM Agent Error Feedback Expert Trajectory Generalizing Clinical Trial MULTI Agent Constrained Exploration

December 27, 2023

Adaptive trajectory-constrained exploration strategy for deep reinforcement learning
Guojian Wang, Faguo Wu, Xiao Zhang, Ning Guo, Zhiming Zheng
Deep Reinforcement Learning Multi Agent Reinforcement Learning Adaptive Importance Exploration Problem Trajectory Wise Reward Adaptive Exploration Constrained Exploration

October 27, 2023

Learning to Search Feasible and Infeasible Regions of Routing Problems with Flexible Neural k-Opt
Yining Ma, Zhiguang Cao, Yeow Meng Chee
Reinforcement Learning Neural Network LeArning Abstract Search Query Routing Problem Neural Solver Learning Search Constrained Exploration

September 28, 2023

Comparing Active Learning Performance Driven by Gaussian Processes or Bayesian Neural Networks for Constrained Trajectory Exploration
Sapphira Akins, Frances Zhu
Active Learning Gaussian Process Bayesian Neural Network Autonomous Robot Active Learning Algorithm Exploration Strategy Intelligent Exploration Constrained Exploration

April 5, 2023

Constrained Exploration in Reinforcement Learning with Optimality Preservation
Peter C. Y. Chen
Reinforcement Learning Action Space Near Optimality Reinforcement Learning System Constrained Exploration

Constrained Exploration

Papers

Constrained Exploration via Reflected Replica Exchange Stochastic Gradient Langevin Dynamics

Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents

Adaptive trajectory-constrained exploration strategy for deep reinforcement learning

Learning to Search Feasible and Infeasible Regions of Routing Problems with Flexible Neural k-Opt

Comparing Active Learning Performance Driven by Gaussian Processes or Bayesian Neural Networks for Constrained Trajectory Exploration

Constrained Exploration in Reinforcement Learning with Optimality Preservation