基于自注意力机制的FCM++及其在学生评价中的应用

FCM++Based on Self-attention Mechanism and its Application in Student Evaluation

利用传统的专家赋权等方式对学生进行评价时,结果往往缺乏准确性。提出一种基于自注意力机制的模糊C均值聚类(FCM)算法,以注意力作为初始聚类中心的选择依据,通过引入注意力机制增强数据之间的关联性,并通过模糊C均值聚类的隶属度思想增强评价的客观性和准确性。实验结果表明,在学生评价问题中,相较于传统模糊聚类算法,提出的引入自注意力机制的FCM++算法在簇间密度和簇内方差等指标上表现更优;相较于基于粒子群的模糊聚类算法,DB指数降低了19%,Dunn指数提高了26%。

Traditional methods such as expert empowerment lack accuracy in evaluating students.This paper proposes a fuzzy C-means(FCM)algorithm based on self-attention mechanism,which uses attention as the basis for selecting the initial clustering center,aiming to enhance the correlation between data by introducing the attention mechanism,and enhance the objectivity and accuracy of the evaluation through the idea of membership degree of fuzzy C-means.Experimental results show that in the student evaluation problem,compared to the traditional fuzzy clustering algorithm,the FCM++algorithm introduced in this paper that introduces the self-attention mechanism performs better in terms of inter-cluster density and intra-cluster variance.Compared with the particle swarm-based fuzzy clustering algorithm,the algorithm proposed in this article reduces the DB index by 19%,and the Dunn index increases by 26%.