On the Variational Posterior of Dirichlet Process Deep Latent Gaussian Mixture Models

Echraibi, Amine; Flocon-Cholet, Joachim; Gosselin, Stéphane; Vaton, Sandrine

doi:10.48550/arxiv.2006.08993

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…deep nonlinear reward functions in IRL), approximates methods may not be able to scale with high dimensional parameter spaces. MCMC sampling methods are shown to be slow in convergence [7,29] and variational inference algorithms suffer from restrictions in the distribution family of the observable data, as well as various truncation assumptions for the variational distribution to yield a finite dimensional representation [11,23]. These limitations apparently make approximate Bayesian inference methods inapplicable for DPM models with deep neural networks.…”

Section: Introductionmentioning

confidence: 99%

Deep Adaptive Multi-Intention Inverse Reinforcement Learning

Bighashdel¹,

Meletis²,

Jancura³

et al. 2021

Preprint

View full text Add to dashboard Cite

This paper presents a deep Inverse Reinforcement Learning (IRL) framework that can learn an a priori unknown number of nonlinear reward functions from unlabeled experts' demonstrations. For this purpose, we employ the tools from Dirichlet processes and propose an adaptive approach to simultaneously account for both complex and unknown number of reward functions. Using the conditional maximum entropy principle, we model the experts' multi-intention behaviors as a mixture of latent intention distributions and derive two algorithms to estimate the parameters of the deep reward network along with the number of experts' intentions from unlabeled demonstrations. The proposed algorithms are evaluated on three benchmarks, two of which have been specifically extended in this study for multi-intention IRL, and compared with well-known baselines. We demonstrate through several experiments the advantages of our algorithms over the existing approaches and the benefits of online inferring, rather than fixing beforehand, the number of expert's intentions.

show abstract

Section: Introductionmentioning

confidence: 99%