Next-generation Artificial Intelligence and Semiotics

Abstract

We still have huge challenges in artificial intelligence and robotics when we attempt to introduce autonomous robots into our daily environment, e.g., home and offices. Machine learning methods, especially deep learning, have already achieved huge success in many pattern recognition tasks that are carefully prepared and in which each system are expected to learn a global rule. However, each robot has to adapt to each local environment not only physically but also semantically via interaction with people living there. We need to invent more adaptive intelligence based on unsupervised learning. In this talk, I introduce the R-GIRO project called “International and Interdisciplinary Research Center for the Next-generation Artificial Intelligence and Semiotics” funded by Ritsumeikan University, and talk about future challenges in this field.