Masahiro Suzuki received Special Paper Award from Information Processing Society of Japan.
Bibliographic Information]
Masahiro Suzuki and Yutaka Matsuo: A deep generative model for bidirectional generation between different modalities, Transactions of Information Processing Society of Japan, Vol. 59, No. 3 (2018)
Abstract]
In this study, we proposed a deep generative model for bidirectional generation between different types of information (modalities), such as image information and tag information. Conventional bi-directional generation methods have problems in generating large information from small dimensional information, such as image information from tag information, but the improved method proposed in this study solves this problem and shows that proper bi-directional generation can be achieved.
