Human Like RL

Human-like reinforcement learning (RL) aims to develop RL agents that exhibit natural, efficient, and safe behaviors comparable to humans, addressing limitations of current methods that often produce unnatural or unsafe actions. Research focuses on improving sample efficiency through techniques like meta-RL and leveraging model architectures such as transformers and state-space layers for better generalization and long-term planning, while also incorporating human feedback and constraints to guide learning towards desired behaviors. This field is significant because it promises more robust and reliable RL agents for real-world applications, particularly in robotics and human-computer interaction, where safety and natural interaction are paramount.

Papers

October 26, 2024

Overcoming the Sim-to-Real Gap: Leveraging Simulation to Learn to Explore for Real-World RL
Andrew Wagenmaker, Kevin Huang, Liyiming Ke, Byron Boots, Kevin Jamieson, Abhishek Gupta
Reinforcement Learning Simulation Study Efficient Exploration Human Like RL Real World Reinforcement Learning

October 8, 2024

RL, but don't do anything I wouldn't do
Michael K. Cohen, Marcus Hutter, Yoshua Bengio, Stuart Russell
Reinforcement Learning Reinforcement Learning Agent Human Like RL RL Agent

September 24, 2024

Context-Based Meta Reinforcement Learning for Robust and Adaptable Peg-in-Hole Assembly Tasks
Ahmed Shokry, Walid Gomaa, Tobias Zaenker, Murad Dawood, Shady A. Maged, Mohammed I. Awad, Maren Bennewitz
Robust Version Meta Reinforcement Learning Human Like RL Peg in Hole Assembly Sensor Equipped Robot

June 28, 2024

PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators
Kuo-Hao Zeng, Zichen Zhang, Kiana Ehsani, Rose Hendrix, Jordi Salvador, Alvaro Herrasti, Ross Girshick, Aniruddha Kembhavi, Luca Weihs
Reinforcement Learning Human Like RL Navigation Agent Transformer Based Policy Indoor Robot Navigation Multi Object Navigation

June 13, 2024

Hadamard Representations: Augmenting Hyperbolic Tangents in RL
Jacob E. Kooi, Mark Hoogendoorn, Vincent François-Lavet
Neural Network Architecture Fast Learning Human Like RL Differentiable Activation Function

June 3, 2024

May 29, 2024

OMPO: A Unified Framework for RL under Policy and Dynamics Shifts
Yu Luo, Tianying Ji, Fuchun Sun, Jianwei Zhang, Huazhe Xu, Xianyuan Zhan
Unified Framework Health Policy Recommendation Human Like RL Reinforcement Learning Policy Dynamic Shift Occupancy Model

May 22, 2024

Theoretical Analysis of Meta Reinforcement Learning: Generalization Bounds and Convergence Guarantees
Cangqing Wang, Mingxiu Sui, Dan Sun, Zecheng Zhang, Yan Zhou
Generalization Bound Theoretical Analysis Convergence Guarantee Meta Reinforcement Learning Human Like RL Meta Rl Algorithm Meta RL

March 27, 2024

Rejection Improves Reliability: Training LLMs to Refuse Unknown Questions Using RL from Knowledge Feedback
Hongshen Xu, Zichen Zhu, Situo Zhang, Da Ma, Shuai Fan, Lu Chen, Kai Yu
Reward Model Human Like RL Risky Prompt Rejection Knowledge Sharing Model Reliability Unknown Question Knowledge Boundary

March 20, 2024

POLICEd RL: Learning Closed-Loop Robot Control Policies with Provable Satisfaction of Hard Constraints
Jean-Baptiste Bouvier, Kartik Nagpal, Negar Mehr
Reinforcement Learning Algorithm Constrained Markov Decision Process Human Like RL Hard Constraint State Constraint

March 18, 2024

RL in Markov Games with Independent Function Approximation: Improved Sample Complexity Bound under the Local Access Model
Junyi Fan, Yuxuan Han, Jialin Zeng, Jian-Feng Cai, Yang Wang, Yang Xiang, Jiheng Zhang
Action Space Markov Game Policy Iteration Human Like RL Sample Complexity Bound General Sum Markov Game Local Client Model

March 11, 2024

Sim-to-Real gap in RL: Use Case with TIAGo and Isaac Sim/Gym
Jaume Albardaner, Alberto San Miguel, Néstor García, Magí Dalmau-Moreno
Robotic Manipulation Collision Free High Fidelity Mobile Manipulator Use Case Human Like RL MR iNet Gym

March 10, 2024

Risk-Sensitive RL with Optimized Certainty Equivalents via Reduction to Standard RL
Kaiwen Wang, Dawen Liang, Nathan Kallus, Wen Sun
Policy OpTimization Risk Sensitive Meta Reinforcement Learning Constructive Reduction Human Like RL Risk Sensitive Reinforcement Learning Certainty Equivalent

November 9, 2023

Zero-Shot Goal-Directed Dialogue via RL on Imagined Conversations
Joey Hong, Sergey Levine, Anca Dragan
Goal Oriented Human Like RL

October 23, 2023

Making RL with Preference-based Feedback Efficient via Randomization
Runzhe Wu, Wen Sun
Reinforcement Learning Active Learning Regret Minimization Human Like RL Linear MDPs Randomization Program User Feedback

October 9, 2023

Reinforcement Learning in the Era of LLMs: What is Essential? What is needed? An RL Perspective on RLHF, Prompting, and Beyond
Hao Sun
Large Language Model Reinforcement Learning Imitation Learning Inverse Reinforcement Learning Policy Learning Reinforcement Learning From Human Feedback Human Like RL RL Method

September 27, 2023

Towards Human-Like RL: Taming Non-Naturalistic Behavior in Deep RL via Adaptive Behavioral Costs in 3D Games
Kuo-Hao Ho, Ping-Chun Hsieh, Chiu-Chou Lin, You-Ren Luo, Feng-Jian Wang, I-Chen Wu
Reinforcement Learning 3D Content Policy OpTimization Deep Reinforcement Deep RL Human Like RL Agent Welfare

August 15, 2023

On-demand Cold Start Frequency Reduction with Off-Policy Reinforcement Learning in Serverless Computing
Siddharth Agarwal, Maria A. Rodriguez, Rajkumar Buyya
Reinforcement Learning Potential Scalability Human Like RL Serverless Computing Cold Start Driven Initialization

June 28, 2023

RL$^3$: Boosting Meta Reinforcement Learning via RL inside RL$^2$
Abhinav Bhatia, Samer B. Nashed, Shlomo Zilberstein
Data Efficient Meta Reinforcement Learning Human Like RL Meta RL

Human Like RL

Papers

Overcoming the Sim-to-Real Gap: Leveraging Simulation to Learn to Explore for Real-World RL

RL, but don't do anything I wouldn't do

Context-Based Meta Reinforcement Learning for Robust and Adaptable Peg-in-Hole Assembly Tasks

PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators

Hadamard Representations: Augmenting Hyperbolic Tangents in RL

RL in Latent MDPs is Tractable: Online Guarantees via Off-Policy Evaluation

Using RL to Identify Divisive Perspectives Improves LLMs Abilities to Identify Communities on Social Media

OMPO: A Unified Framework for RL under Policy and Dynamics Shifts

Theoretical Analysis of Meta Reinforcement Learning: Generalization Bounds and Convergence Guarantees

Rejection Improves Reliability: Training LLMs to Refuse Unknown Questions Using RL from Knowledge Feedback

POLICEd RL: Learning Closed-Loop Robot Control Policies with Provable Satisfaction of Hard Constraints

RL in Markov Games with Independent Function Approximation: Improved Sample Complexity Bound under the Local Access Model

Sim-to-Real gap in RL: Use Case with TIAGo and Isaac Sim/Gym

Risk-Sensitive RL with Optimized Certainty Equivalents via Reduction to Standard RL

Zero-Shot Goal-Directed Dialogue via RL on Imagined Conversations

Making RL with Preference-based Feedback Efficient via Randomization

Reinforcement Learning in the Era of LLMs: What is Essential? What is needed? An RL Perspective on RLHF, Prompting, and Beyond

Towards Human-Like RL: Taming Non-Naturalistic Behavior in Deep RL via Adaptive Behavioral Costs in 3D Games

On-demand Cold Start Frequency Reduction with Off-Policy Reinforcement Learning in Serverless Computing

RL$^3$: Boosting Meta Reinforcement Learning via RL inside RL$^2$