Learning Reward

Learning reward functions in reinforcement learning aims to automatically define what constitutes desirable behavior for an AI agent, eliminating the need for manual reward engineering. Current research focuses on methods that learn rewards from various sources, including expert demonstrations (e.g., using diffusion models or ELO ratings), human feedback (incorporating models of human rationality), and even implicitly through guiding agents towards global performance metrics. These advancements are crucial for building more robust and reliable AI systems, particularly in complex, real-world scenarios where manually designing reward functions is impractical or impossible.

Papers

September 5, 2024

ELO-Rated Sequence Rewards: Advancing Reinforcement Learning Models
Qi Ju, Falin Hei, Zhemei Fang, Yunfeng Luo
Reinforcement Learning Deep Learning Reinforcement Learning Model Utility Based Learning Reward Reward Redistribution

August 16, 2024

Neural Reward Machines
Elena Umili, Francesco Argenziano, Roberto Capobianco
Non Markovian Neurosymbolic Framework Learning Reward

February 7, 2024

Explaining Learned Reward Functions with Counterfactual Trajectories
Jan Wehner, Frans Oliehoek, Luciano Cavalcante Siebert
Line by Line Explanation Reward Function Counterfactual Sequence Learning Reward

December 21, 2023

Diffusion Reward: Learning Rewards via Conditional Video Diffusion
Tao Huang, Guangqi Jiang, Yanjie Ze, Huazhe Xu
Diffusion Model Sparse Reward Reward Report Reinforcement Learning Task Learning Reward

May 3, 2023

Reward Teaching for Federated Multi-armed Bandits
Chengshuai Shi, Wei Xiong, Cong Shen, Jing Yang
Multi Armed Bandit Epsilon Greedy Federated Bandit Learning Reward

April 13, 2023

Power-seeking can be probable and predictive for trained agents
Victoria Krakovna, Janos Kramar
Agent Smith Reward Function Advanced AI Training Scheme Power Seeking Learning Reward

March 16, 2023

Learning Rewards to Optimize Global Performance Metrics in Deep Reinforcement Learning
Junqi Qian, Paul Weng, Chenmien Tan
Reinforcement Learning Deep Reinforcement Learning Performance Metric Reward Engineering Learning Reward

August 23, 2022

The Effect of Modeling Human Rationality Level on Learning Rewards from Multiple Feedback Types
Gaurav R. Ghosal, Matthew Zurek, Daniel S. Brown, Anca D. Dragan
Mixed Effect Reward Learning Human Rationality Noisy Preference Multi Point Feedback Learning Reward