Simultaneous Learning of Object Concepts, Language Model, and Acoustic Model Using Unsupervised Multimodal Learning

Abstract

Categorization of objects plays an important role in human cognition. It is important for robots to form the object concepts for communication with humans. The purpose is to enable that robots form such object categories and acquire language. We propose the simultaneous learning of the object concepts, the language model, and the acoustic model using by combining Multimodal Latent Dirichlet Allocation (MLDA) with Nonparametric Bayesian Double Articulation Analyzer (NPB-DAA).