Self-initiated Communication Learning Mechanism by Integrating Situation, Image, and Language

点击翻译按钮获取中文摘要

This research addresses fundamental aspects of a self-initiated mechanism for learning verbal-nonverbal communication integrating situations, images, and language. The result consists of the architecture for empathic agents, a method for extracting verbal-nonverbal associations from multi-modal interaction data and its application to creating agent communication behavior patterns, learning instruction patterns and task structure by incremental acquisition of behavior rules, and algorithms for self-initiated learning by mimicking.