Heterogeneous Bagging

Heterogeneous bagging addresses the challenge of efficiently learning from diverse datasets or agents exhibiting varying characteristics, a common problem in machine learning and robotics. Current research focuses on developing algorithms, such as weighted least squares estimators and elimination-based methods, that effectively leverage shared information while accounting for task-specific differences, often within the frameworks of multi-task learning and federated learning for bandits. This work is significant for improving the efficiency and robustness of learning systems in scenarios with heterogeneous data sources, impacting fields ranging from personalized recommendations to collaborative robotics.

Papers

June 9, 2023

Optimal Multitask Linear Regression and Contextual Bandits under Sparse Heterogeneity
Xinmeng Huang, Kan Xu, Donghwan Lee, Hamed Hassani, Hamsa Bastani, Edgar Dobriban
Contextual Bandit State Estimation Heterogeneous Learning Heterogeneous Bagging

March 29, 2023

Federated Learning for Heterogeneous Bandits with Unobserved Contexts
Jiabin Lin, Shana Moothedath
Stochastic Way Optimal Action Contextual Multi Armed Bandit Unobserved Variable Heterogeneous Bagging

March 9, 2023

Flooding with Absorption: An Efficient Protocol for Heterogeneous Bandits over Complex Networks
Junghyun Lee, Laura Schmid, Se-Young Yun
Multi Armed Bandit Sequential Decision Complex Network Regret Analysis Stable Regret Absorption Coefficient Efficient Protocol Heterogeneous Bagging

October 18, 2022

Bag All You Need: Learning a Generalizable Bagging Strategy for Heterogeneous Objects
Arpit Bahety, Shreeya Jain, Huy Ha, Nathalie Hager, Benjamin Burchfiel, Eric Cousineau, Siyuan Feng, Shuran Song
Deformable Object Bag Prototype Heterogeneous Bagging

May 23, 2022

Computationally Efficient Horizon-Free Reinforcement Learning for Linear Mixture MDPs
Dongruo Zhou, Quanquan Gu
Contextual Bandit Episodic Reinforcement Learning Linear Mixture Markov Decision Process Heterogeneous Bagging

Heterogeneous Bagging

Papers

Optimal Multitask Linear Regression and Contextual Bandits under Sparse Heterogeneity

Federated Learning for Heterogeneous Bandits with Unobserved Contexts

Flooding with Absorption: An Efficient Protocol for Heterogeneous Bandits over Complex Networks

Bag All You Need: Learning a Generalizable Bagging Strategy for Heterogeneous Objects

Computationally Efficient Horizon-Free Reinforcement Learning for Linear Mixture MDPs