Making a Machine That Sees Like Us

New issue of The HSE LooK is now available for read and download! Our Octrober edition is devoted to the new Faculty of Social Sciences, its departments, projects and people. Tadamasa Sawada, Assistant Professor at the School of Psychology, spoke to The HSE LooK about his research. We wish our new colleague a great beginning at the HSE!

What is your research about?

— I have been studying human visual perception using psychophysical experiments and computational mathematical modeling, 3D perception in particular. The world out there is 3D, its image on a retina is 2D, and our perception is 3D. The human visual system is recovering the third dimension from the retinal image. This is so called an ill-posed problem. This is one of the most amazing aspects of the brain.

— What is the practical application of your work?

— It can be applied for developing robot vision as well as supporting human vision. Human-robot interaction is getting more and more important these days for helping elderly and disabled people and making jobs currently performed by humans easier.

A book written by you and your colleagues recently emerged from the Oxford University press. What are your impressions from this collaborative work?

— There are good aspects in publishing books, but for me this format is a little outdated. This format has existed for 4,000 years! Another problem is language as the means of transferring knowledge. Any information expressed by words is always limited in terms of the context load. If we read some scientific papers written a century ago they are often hard to understand, because people who wrote them lived in another context. In the case of 3D perception, if we can visualize problems with the help of interactive demos then it becomes much clearer. Writing texts is important but there is always some additional media available. If you can present your idea better by using movies, or interactive demos, you should do it. Recently, when I was submitting a paper to be published, reviewers asked me to make some demos. People really like this kind of visualization because it facilitates their understanding of complex problems.

Tadamasa Sawada grew up and studied in Japan. In 2006 he defended his PhD dissertation at the Tokyo Institute of Technology and shortly after that he went to the U.S. where he worked as a postdoctoral research fellow at Purdue University in West Lafayette, The Ohio State University, and the State University of New York. Sawada is a co-author of the book Making a Machine That Sees Like Us, which was released by Oxford University Press in 2014. Since September 2014 Sawada has worked as an Assistant Professor at the HSE Faculty of Social Sciences. In his free time in Moscow Sawada enjoys walks around the city, visits museums and admires historical architecture.

The full text of the issue including interviews with international faculty can be found in Issue 8 (15), October 2014.