Imitation Learning Applied to Embodied Conversational Agents

Abstract : Embodied Conversational Agents (ECAs) are emerging as a key component to allow human interact with machines. Applications are numerous and ECAs can reduce the aversion to interact with a machine by providing user-friendly interfaces. Yet, ECAs are still unable to produce social signals appropriately during their interaction with humans, which tends to make the interaction less instinctive. Especially, very little attention has been paid to the use of laughter in human-avatar interactions despite the crucial role played by laughter in human-human interaction. In this paper, methods for predicting when and how to laugh during an interaction for an ECA are proposed. Different Imitation Learning (also known as Apprenticeship Learning) algorithms are used in this purpose and a regularized classification algorithm is shown to produce good behavior on real data.
Complete list of metadatas

Cited literature [25 references]  Display  Hide  Download

https://hal.inria.fr/hal-01225816
Contributor : Olivier Pietquin <>
Submitted on : Monday, November 9, 2015 - 2:58:35 PM
Last modification on : Wednesday, July 31, 2019 - 4:18:02 PM
Long-term archiving on : Wednesday, February 10, 2016 - 10:11:41 AM

File

piot15.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01225816, version 1

Citation

Bilal Piot, Matthieu Geist, Olivier Pietquin. Imitation Learning Applied to Embodied Conversational Agents. 4th Workshop on Machine Learning for Interactive Systems (MLIS 2015), Jul 2015, Lille, France. ⟨hal-01225816⟩

Share

Metrics

Record views

353

Files downloads

128