Generative Adversarial Reward Learning for Generalized Behavior Tendency Inference
This addresses the need for adaptive and generalizable reward functions in dynamic environments like recommender systems, though it appears incremental as it builds on existing inverse reinforcement learning and GAN techniques.
The paper tackles the problem of manually-defined reward functions in reinforcement learning by proposing a generative inverse reinforcement learning model that automatically learns rewards from user actions, and it shows outperformance over state-of-the-art methods in traffic signal control, online recommender systems, and scanpath prediction.
Recent advances in reinforcement learning have inspired increasing interest in learning user modeling adaptively through dynamic interactions, e.g., in reinforcement learning based recommender systems. Reward function is crucial for most of reinforcement learning applications as it can provide the guideline about the optimization. However, current reinforcement-learning-based methods rely on manually-defined reward functions, which cannot adapt to dynamic and noisy environments. Besides, they generally use task-specific reward functions that sacrifice generalization ability. We propose a generative inverse reinforcement learning for user behavioral preference modelling, to address the above issues. Instead of using predefined reward functions, our model can automatically learn the rewards from user's actions based on discriminative actor-critic network and Wasserstein GAN. Our model provides a general way of characterizing and explaining underlying behavioral tendencies, and our experiments show our method outperforms state-of-the-art methods in a variety of scenarios, namely traffic signal control, online recommender systems, and scanpath prediction.