PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training

By Kimin Lee, Laura M. Smith, Pieter Abbeel. 2021

408 citations recorded by the source corpus.

Conveying complex objectives to reinforcement learning (RL) agents can often be difficult, involving meticulous design of reward functions that are sufficiently informative yet easy enough to provide. Human-in-the-loop RL methods allow practitioners to instead interactively teach agents through tailored feedback; however, such approaches have been challenging to scale since human feedback is very expensive. In this work, we aim to make this process more sample- and feedback-efficient. We present an off-policy, interactive RL algorithm that capitalizes on the strengths of both feedback and off-policy learning. Specifically, we learn a reward model by actively querying a teacher's preferences between two clips of behavior and use it to train an agent. To enable off-policy learning, we relabel all the agent's past experience when its reward model changes. We additionally show that pre-training our agents with unsupervised exploration substantially increases the mileage of its queries. We demonstrate that our approach is capable of learning tasks of higher complexity than previously considered by human-in-the-loop methods, including a variety of locomotion and robotic manipulation skills. We also show that our method is able to utilize real-time human feedback to effectively prevent reward exploitation and learn new behaviors that are difficult to specify with standard reward functions.

Subfields: Artificial Intelligence, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

Citations in the corpus

Widening the Pipeline in Human-Guided Reinforcement Learning with Explanation and Context-Aware Data Augmentation (2020, background): “Leveraging human guidance for RL tasks has been extensively studied in the context of imitation learning (learning from demonstration) [37], inverse reinforcement learning [29, 1], reward shaping [28], learning from human preference [7, 22], and learning from saliency information provided by human [”
Aligning Text-to-Image Models using Human Feedback (2023, background): “…RL agents without the need for hand-designed rewards (MacGlashan et al., 2017; Christiano et al., 2017; Warnell et al., 2018; Ibarz et al., 2018; Lee et al., 2021, inter alia), and to more truthful and harmless instruction following and dialogue (Ouyang et al., 2022; Bai et al., 2022a; Ziegler…”
DPOK: Reinforcement Learning for Fine-tuning Text-to-Image Diffusion Models (2023, background): “Human assessments of (or preferences over) learned model outcomes have been used to guide learning on a variety of tasks, ranging from learning behaviors [16] to language modeling [1, 23, 19, 41].”
Preference Ranking Optimization for Human Alignment (2023, background): “Many explorations have been done on this path (Chris-tiano et al., 2017; MacGlashan et al., 2017; Warnell et al., 2018; Ziegler et al., 2019; Stiennon et al., 2020b; Nakano et al., 2021; Lee et al., 2021; Lei et al., 2022; Snell et al., 2022; Bai et al., 2022a; Ouyang et al., 2022a).”
ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generation (2023, methodology): “Thus human feedback has been utilized to align model performance with intent in various language applications [1, 20, 33, 59] via training an RM [32, 8, 53, 19, 22] to learn human preference.”
Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback (2023, background): “These advantages have made RLHF useful for helping policies learn intricate solutions in control environments [Christiano et al., 2017, Biyik, 2022, Lee et al., 2021, Hejna and Sadigh, 2022] and for finetuning LLMs [Bai et al., 2022a, Ziegler et al., 2019, Stiennon et al., 2020].”