Score Fusion Method by Neural Network Using GPS and Wi-Fi Log Data

Recently, personal authentication has become more important than ever because of authentication vulnerability problems. In cases where extremely high confidentiality is required, multi-factor authentication is used. In this study, we focus on score fusion method, which merges authentication scores of each factor in multi-factor authentication. In conventional score fusion methods, the weight of each factor is fixed. These methods work well when all users have a similar tendency for each factor. However, behavioral authentication is highly dependent on each user's characteristics, and the tendency of which factors have high accuracy may differ greatly among users. There is no score fusion method suitable for such a case where the tendency of each factor is different. We propose a user dependent weighting score fusion method using a neural network. Since our proposed method constructs a simple binary classification network for each user, the weight is personally adjusted for each user. The comparison experiment results show that True Acceptance Rate (TAR) of our proposed method is higher than conventional methods.