Computer Science Program

Yuko OSANA

Name

Yuko OSANA

Department/Research Field

Computer Science/Soft Computing


Research content
We are studying the development of systems using techniques such as neural networks and genetic algorithms and learning methods of new neural networks. In similarity-based image retrieval, we realize not only retrieval for landscape images but also retrieval considering similarity of illustration style (touch). In addition, we are studying learning methods without teacher signal by combining reinforcement learning and neural network. In these studies, not only experiments by simulation but also learning using an actual robot are realized. In research using genetic algorithms, we are developing a system that creates automatic composition and office layout just by inputting some conditions.
Research theme

1. similarity-based image retrieval using neural network
2. learning of robots and games by reinforcement learning using neural network
3. design support using genetic algorithm (automatic composition / office layout)

Research keywords
Neural Network , Reinforcement Learning , Genetic Algorithm
Lab link