Topological Intelligence and Topological Twin

Abstract

Recently, various concepts on cyber-physical systems and digital twin have been proposed and discussed with the integration of information, intelligence, communication, and robot technologies. We often have to extract topological features and structures from given or measured big data to simulate a real-world phenomenon in the cyber world and to conduct multiscale and multiphysics simulations. Therefore, we proposed the concept of topological twin. The aim of topological twin is to (1) extract topological structures hidden implicitly in the real world, (2) reproduce them explicitly in the cyber world, and (3) simulate and analyze the real world in the cyber world. While we have to deal with the physical dynamics in the microscopic level, we have to deal with spatiotemporal qualitative relationships between objects, people, culture, and knowledge in the macroscopic level. Furthermore, we need a mesoscopic integration method connecting microscopic and macroscopic topological features. The topological twin plays the important role in extracting and connecting structures hidden in real world from the mutliscopic point of view. We can extract topological features and structures from big data, that are used as topological big data in different level of analysis. Furthermore, we need a multiscopic approach to deal with inference, learning, search, and prediction based on topological and graphical data as the methodology of topological intelligence. In this talk, first, we introduce the concept of multiscopic topological twin. Next, I explain various types of topological mapping methods, unsupervised learning methods, and graph-based methods related with topological intelligence. One of them is Growing Neural Gas (GNG) that can dynamically change the topological structure composed of nodes and edges. One important advantage of GNG is in the incremental learning capability of nodes and edges according to a target data distribution, but the computational cost of standard GNG is very expensive. Therefore, we proposed a method of multi-scale batch-learning GNG called Fast GNG. Next, we show the comparison result of Fast GNG with other methods. Furthermore, we show several experimental results of topological intelligence in trailer living laboratory, robot partners and mobility support robots. Finally, we discuss the applicability and future direction of multiscopic topological twin.Recently, various concepts on cyber-physical systems and digital twin have been proposed and discussed with the integration of information, intelligence, communication, and robot technologies. We often have to extract topological features and structures from given or measured big data to simulate a real-world phenomenon in the cyber world and to conduct multiscale and multiphysics simulations. Therefore, we proposed the concept of topological twin. The aim of topological twin is to (1) extract topological structures hidden implicitly in the real world, (2) reproduce them explicitly in the cyber world, and (3) simulate and analyze the real world in the cyber world. While we have to deal with the physical dynamics in the microscopic level, we have to deal with spatiotemporal qualitative relationships between objects, people, culture, and knowledge in the macroscopic level. Furthermore, we need a mesoscopic integration method connecting microscopic and macroscopic topological features. The topological twin plays the important role in extracting and connecting structures hidden in real world from the mutliscopic point of view. We can extract topological features and structures from big data, that are used as topological big data in different level of analysis. Furthermore, we need a multiscopic approach to deal with inference, learning, search, and prediction based on topological and graphical data as the methodology of topological intelligence. In this talk, first, we introduce the concept of multiscopic topological twin. Next, I explain various types of topological mapping methods, unsupervised learning methods, and graph-based methods related with topological intelligence. One of them is Growing Neural Gas (GNG) that can dynamically change the topological structure composed of nodes and edges. One important advantage of GNG is in the incremental learning capability of nodes and edges according to a target data distribution, but the computational cost of standard GNG is very expensive. Therefore, we proposed a method of multi-scale batch-learning GNG called Fast GNG. Next, we show the comparison result of Fast GNG with other methods. Furthermore, we show several experimental results of topological intelligence in trailer living laboratory, robot partners and mobility support robots. Finally, we discuss the applicability and future direction of multiscopic topological twin.

Biography :

Naoyuki Kubota is currently a Professor in the Department of Mechanical Systems Engineering, the Graduate School of Systems Design, and Director of Community-centric System Research Core, Tokyo Metropolitan University, Japan. He graduated from Osaka Kyoiku University in 1992, received the M.E. degree from Hokkaido University in 1994, and received the D.E. from Nagoya University, Nagoya, Japan, in 1997. He was an Assistant Professor and Lecturer at the Department of Mechanical Engineering, Osaka Institute of Technology, Japan, from 1997 to 2000. In 2000, he joined the Department of Human and Artificial Intelligence Systems, the School of Engineering, Fukui University, Japan, as an Associate Professor. He joined the Department of Mechanical Engineering, the Graduate School of Engineering, Tokyo Metropolitan University, Japan, as an Associate Professor in 2004. He was an Associate Professor from 2005 to 2012, and a Professor from 2012 at the Graduate School of Systems Design, Tokyo Metropolitan University, Japan. He was a Visiting Professor at University of Portsmouth, UK, in 2007 and 2009, and was an Invited Visiting Professor at Seoul National University from 2009 to 2012, and others. His current interests are in the fields of topological mapping, coevolutionary computation, spiking neural networks, perception-based robotics, robot partners, and informationally structured space. He has published more than 500 refereed journal and conference papers in the above research fields. He received the Best Paper Award of IEEE IECON 1996, IEEE CIRA 1997, MHS 2011, WAC 2012, HSI 2016, and so on. He was an associate editor of the IEEE Transactions on Fuzzy Systems from 1999 to 2010, the IEEE CIS Intelligent Systems Applications Technical Committee, Robotics Task Force Chair from 2007 to 2014, IEEE Systems, Man, and Cybernetics Society, Japan Chapter Chair since 2018, Vice Director, Tokyo Biomarker Innovation Research Association, Japan from 2020, and others.