Latent Wasserstein Adversarial Imitation Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yang, Siqi, Yan, Kai, Schwing, Alexander G., Wang, Yu-Xiong
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911489413087232
author Yang, Siqi
Yan, Kai
Schwing, Alexander G.
Wang, Yu-Xiong
author_facet Yang, Siqi
Yan, Kai
Schwing, Alexander G.
Wang, Yu-Xiong
contents Imitation Learning (IL) enables agents to mimic expert behavior by learning from demonstrations. However, traditional IL methods require large amounts of medium-to-high-quality demonstrations as well as actions of expert demonstrations, both of which are often unavailable. To reduce this need, we propose Latent Wasserstein Adversarial Imitation Learning (LWAIL), a novel adversarial imitation learning framework that focuses on state-only distribution matching. It benefits from the Wasserstein distance computed in a dynamics-aware latent space. This dynamics-aware latent space differs from prior work and is obtained via a pre-training stage, where we train the Intention Conditioned Value Function (ICVF) to capture a dynamics-aware structure of the state space using a small set of randomly generated state-only data. We show that this enhances the policy's understanding of state transitions, enabling the learning process to use only one or a few state-only expert episodes to achieve expert-level performance. Through experiments on multiple MuJoCo environments, we demonstrate that our method outperforms prior Wasserstein-based IL methods and prior adversarial IL methods, achieving better results across various tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2603_05440
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Latent Wasserstein Adversarial Imitation Learning
Yang, Siqi
Yan, Kai
Schwing, Alexander G.
Wang, Yu-Xiong
Machine Learning
Imitation Learning (IL) enables agents to mimic expert behavior by learning from demonstrations. However, traditional IL methods require large amounts of medium-to-high-quality demonstrations as well as actions of expert demonstrations, both of which are often unavailable. To reduce this need, we propose Latent Wasserstein Adversarial Imitation Learning (LWAIL), a novel adversarial imitation learning framework that focuses on state-only distribution matching. It benefits from the Wasserstein distance computed in a dynamics-aware latent space. This dynamics-aware latent space differs from prior work and is obtained via a pre-training stage, where we train the Intention Conditioned Value Function (ICVF) to capture a dynamics-aware structure of the state space using a small set of randomly generated state-only data. We show that this enhances the policy's understanding of state transitions, enabling the learning process to use only one or a few state-only expert episodes to achieve expert-level performance. Through experiments on multiple MuJoCo environments, we demonstrate that our method outperforms prior Wasserstein-based IL methods and prior adversarial IL methods, achieving better results across various tasks.
title Latent Wasserstein Adversarial Imitation Learning
topic Machine Learning
url https://arxiv.org/abs/2603.05440