Inverse Preference Learning: Preference-based RL without a Reward Function
This addresses the problem of reward misalignment in RL for researchers and practitioners by offering a simpler, more parameter-efficient method, though it is incremental as it builds on existing preference-based RL approaches.
The paper tackles the challenge of designing reward functions in reinforcement learning by introducing Inverse Preference Learning (IPL), which eliminates the need for a learned reward function by using Q-functions to encode reward information, achieving competitive performance on continuous control and robotics benchmarks with fewer parameters.
Reward functions are difficult to design and often hard to align with human intent. Preference-based Reinforcement Learning (RL) algorithms address these problems by learning reward functions from human feedback. However, the majority of preference-based RL methods naïvely combine supervised reward models with off-the-shelf RL algorithms. Contemporary approaches have sought to improve performance and query complexity by using larger and more complex reward architectures such as transformers. Instead of using highly complex architectures, we develop a new and parameter-efficient algorithm, Inverse Preference Learning (IPL), specifically designed for learning from offline preference data. Our key insight is that for a fixed policy, the $Q$-function encodes all information about the reward function, effectively making them interchangeable. Using this insight, we completely eliminate the need for a learned reward function. Our resulting algorithm is simpler and more parameter-efficient. Across a suite of continuous control and robotics benchmarks, IPL attains competitive performance compared to more complex approaches that leverage transformer-based and non-Markovian reward functions while having fewer algorithmic hyperparameters and learned network parameters. Our code is publicly released.