English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We explore methodologies to improve the robustness of generative adversarialimitation learning (GAIL) algorithms to observation noise. Towards thisobjective, we study the effect of local Lipschitzness of the discriminator andthe generator on the robustness of policies learned by GAIL. In many roboticsapplications, the learned policies by GAIL typically suffer from a degradedperformance at test time since the observations from the environment might becorrupted by noise. Hence, robustifying the learned policies against theobservation noise is of critical importance. To this end, we propose aregularization method to induce local Lipschitzness in the generator and thediscriminator of adversarial imitation learning methods. We show that themodified objective leads to learning significantly more robust policies.Moreover, we demonstrate -- both theoretically and experimentally -- thattraining a locally Lipschitz discriminator leads to a locally Lipschitzgenerator, thereby improving the robustness of the resultant policy. We performextensive experiments on simulated robot locomotion environments from theMuJoCo suite that demonstrate the proposed method learns policies thatsignificantly outperform the state-of-the-art generative adversarial imitationlearning algorithm when applied to test scenarios with noise-corruptedobservations.

On the Benefits of Inducing Local Lipschitzness for Robust Generative Adversarial Imitation Learning