基于词频奖惩算法的动态自由击键持续识别

Dynamic Free Text Keystroke Continuous Identification Based on the Rewards and Punishments Algorithm with Frequency

提出了基于词频奖惩算法的动态自由击键持续识别方法,该方法主要包括五个步骤:数据采集、数据预处理、模板建立、训练数据集、认证识别。词频奖惩算法在原有的奖惩算法基础上,调整了距离公式中单双键的权重,加入了词频统计以及对不同用户采用不同的系数值,通过实验证明,本文提出的方法在动态自由击键过程中,能够有效地提高认证识别效率。

In this paper,it is proposed that dynamic free text keystroke continuous identification based on therewards and punishments algorithmwithfrequency,this method contains five steps ： data collection, datapre - treatment, templatesereation, data - set training and certification recognition. On the basis of the previousrewards and punishments algorithm,therewards and punishments algorithm withfre-queney adjusts the weightsof single key and double keys in the distance formula,adds words frequency statistics, and uses different coefficient values for different users, experiments show that, in the study of dynamic free text keystroke, the method in this paper is able to improve the efficiency of recognition and certification.