2018.11.01
In recent years, it has been shown that recognition methods using deep learning can achieve higher recognition accuracy compared to image recognition methods using SVMs, random forests, and the like. For this reason, various image recognition technologies using deep learning are being actively researched. In addition, deep learning can be used not only for image recognition but also for image generation and image transformation.
This lecture will focus on image generation and image transformation, explaining cutting-edge research that includes methods such as GAN, which can generate data similar to a given image; pix2pix, for automatically coloring black-and-white line art images; and CycleGAN, which can transform an image of a white horse into a zebra.
Date: Thursday, November 1, 2018
Time: 18:30–20:00 *Reception to follow from 20:00
Venue: Keio University Yagami Campus (Details will be provided once finalized)
Organizer: Keio University Faculty of Science and Technology
Co-organizer: Keio University Global Research Institute (KGRI) Core Project Creative Cluster
Eligibility: Keio University faculty, staff, and students
Capacity: 60 people
Language: Japanese
Fee: Free *Reception fee: Approx. 500 yen
Pre-registration required: Please register from here .
Registration Deadline: 23:59, Sunday, October 28, 2018
*This colloquium is operated with funds for the enhancement of education at the Faculty of Science and Technology, courtesy of the Mentor Mita-kai.
Inquiries
Katsuhiro Endo, Researcher, Graduate School of Science and Technology
Inquiries regarding this lecture will be accepted in the remarks section of the registration form.