Abstract
Imitation learning is an effective approach for an autonomous agent to learn control policies when an explicit reward function is unavailable, using demonstrations provided from an expert. However, standard imitation learning methods assume that the agents and the demonstrations provided by the expert are in the same domain configuration. Such an assumption has made the learned policies difficult to apply in another distinct domain. The problem is formalized as domain adaptive imitation learning, which is the process of learning how to perform a task optimally in a learner domain, given demonstrations of the task in a distinct expert domain. We address the problem by proposing a model based on Generative Adversarial Network. The model aims to learn both domain-shared and domain-specific features and utilizes it to find an optimal policy across domains. The experimental results show the effectiveness of our model in a number of tasks ranging from low to complex high-dimensional.
Highlights
The demand for autonomous agents capable of mimicking human behaviors has grown significantly in recent years
The problem is formalized as domain adaptive imitation learning, which is a process of learning how to perform a task optimally in a learner domain, given demonstrations of the task in a distinct expert domain [14]
The evaluation results of the proposed DAIL-Generative Adversarial Network (GAN) model on lowand high-dimensional tasks are presented to highlight its superior capability in domain adaptive imitation learning
Summary
The demand for autonomous agents capable of mimicking human behaviors has grown significantly in recent years. In order for autonomous agents to acquire such human complex behaviors, they are supplied with reward functions indicating the goals of the desired behaviors. Humans can learn complex behaviors from imitation: we observe other experts performing the tasks, infer the tasks, attempt to accomplish the same tasks ourselves. Inspired by this learning procedure, imitation learning has been widely used for training autonomous agents using expert-provided demonstrations [1,2,3,4]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have