Understanding of Fashion Items Coordination based on distributed expression

Abstract

Recently, analysis of the fashion outfits has become an attractive research topic in the artificial intelligence fields. However, understanding of fashion is not easy because of this problem including human creativity and Kansei. In this study, we have proposed a machine learning system to solve the problem of fashion outfits composing. In the proposed method, distributed representation is applied to fashion items, and deep learning technique based on LSTM is utilized. The effectiveness of the proposed method is confirmed by computer simulations taking original four-choice questions for fashion items in the Polyvore dataset as examples.