User-based feature selection for gait recognition
Mobile devices such as smartphones have, with the advancement of technology, incorporated sensors that are increasingly being used to collect user data. The accelerometer is one of these sensors, which can be used in biometric systems to recognize an individual's gait. The goal of this work is to investigate how a user-based feature vector can improve the predictive performance of classification algorithms in verification mode. Through different sets of characteristics, this work proposes a system for selecting a set of features for each user. The proposal was evaluated with two classification algorithms and showed that it can improve the average performance of a biometric system for gait recognition.