Corridor of EEIC, The University of Tokyo

Evolutionary Computation and Intelligent :Hitoshi Iba

EmergenceEvolutionary Computation and Metaheuristics

In our laboratory, we study computation and systems with the keywords of evolution and emergence.
・Why are the peacock’s feathers so incredibly beautiful?
・Why did the giraffe’s neck become so long?
・If a worker bee cannot have any offspring of its own, why does it work so hard to serve the queen bee?
We see that biological organisms are solving certain types of optimization problems through the process of evolution. It is the objective of the evolutionary method to exploit this concept to establish an effective computing system (an evolutionary system).

Deep learning + evolutionary computation = deep neuro-evolution

Neuro-evolution is a framework that integrates DL (Deep Learning) and EC (Evolutionary Computation). The main feature of neuro-evolution is that it genetically searches for the optimal network and its learning parameters, thereby eliminating the time and effort (e.g., network construction by trial and error) associated with conventional neural network search.

Machine Learning + Evolutionary Computation = Smarter Optimization

EC (Evolutionary Computation) is integrated with ML (Machine Learning) and is widely applied not only in engineering optimization but also in financial engineering, art and design. Evolutionary reinforcement learning has been applied to robotics and game AI.

Mind Render: A Practical Study of AI Education

We are conducting education and research to teach AI to junior and senior high school students and other beginners in an easy-to-understand manner. For example, we participated in educational training at a high school and conducted empirical verification of the curriculum for AI robots (actual experiments). In addition, Mind Render is a programming learning application that allows users to create and play with VR programs, and it is actually being used in some elementary and junior high schools for classes and free research. In our laboratory, we have developed drills for AI learning using this application, and are providing opportunities for richer information education.

Complex Systems and Artificial Life

The study of complex systems and artificial life is closely related to artificial intelligence. For example, the swarming behavior of ants, bees, and fishes generates complex swarm intelligence that cannot be described by the behavior of each individual. This kind of swarm intelligence is called meta-heuristics, and it has been applied to various fields of artificial intelligence. This research aims to integrate engineering and life sciences, and to realize the main concepts of life phenomena such as “symbiosis” and “diversity” using computers.