Binary PSO Variants for Feature Selection in Handwritten Signature Authentication

Emrah Hancer,Konstantin Sarin,Artem Slezkin,Mikhail Svetlakov,Marina Bardamova,Ilya Hodashinsky

doi:10.15388/21-infor472

Abstract

In this paper we propose modifications of the well-known algorithm of particle swarm optimization (PSO). These changes affect the mapping of the motion of particles from continuous space to binary space for searching in it, which is widely used to solve the problem of feature selection. The modified binary PSO variations were tested on the dataset SVC2004 dedicated to the problem of user authentication based on dynamic features of a handwritten signature. In the example of k-nearest neighbours (kNN), experiments were carried out to find the optimal subset of features. The search for the subset was considered as a multicriteria optimization problem, taking into account the accuracy of the model and the number of features.

Highlights

A handwritten signature is one of the most commonly used instruments of person authenticating
Type I error shows what percentage of legitimate users was mistaken by the classifier for an intruder
The work considered seven approaches for particle swarm optimization (PSO) modification when solving the task of feature selection for a handwritten signature authentication

Summary

Introduction

A handwritten signature is one of the most commonly used instruments of person authenticating. There are two types of handwritten signature recognition: static (offline) and dynamic (online) (Mailah and Han, 2012). Static type consists only in processing a graphic image of the signature. Dynamic recognition involves the analysis of dynamic parameters, such as coordinates, pressure, pen speed, and the total time of the signature. Such dynamic parameters can’t be determined with only the final image of the signature, so the task of falsification is much more complicated. Research has shown that dynamic signature recognition can significantly improve the accuracy of user identification (Linden et al, 2018)

Methods

Results

Conclusion