Evolutionary automatic construction of cellular networks for image processing and recognition

2016 Fiscal Year Final Research Report

• Project/Area Number: 26280056
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Partial Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Yokohama National University
• Principal Investigator: NAGAO TOMOHARU 横浜国立大学, 大学院環境情報研究院, 教授 (10180457)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 知能情報処理 / 進化計算法 / 人工知能 / 機械学習

Outline of Final Research Achievements

The way to perform image processing and recognition using the cellular network which arranged a small-scale circuit which consists of rather simple functions in lattice grid has been researched in this project. In this method, evolutional computation, which is a kind of optimization, was employed to construct the adequate cellular network for a given task automatically. We applied the method to the tasks such as super-resolution, image segmentation and semantic segmentation and so on, and verified the effectiveness of the method. This method has a big advantage that it is possible to construct various processing just by determining a small-scale circuit, and the method is also practical.

Free Research Field

進化計算法