AI-Driven Feature Extraction and Classification Algorithm for Large-Scale High-Dimensional Data
Ping An, Yiding Huang, Chengkuan Liu, Chen Xiao-lon, Lai Zhuoxiang · 2024
In this paper, an AI-driven feature extraction and classification algorithm is studied to solve the challenges in large-scale high-dimensional data processing. The algorithm proposed in this paper uses self-attention mechanism and Transformer network to effectively capture the dependence between the internal features of the sample in the feature extraction stage, and realizes end-to-end classification in the classification stage. Experimental validation conducted on the MNIST dataset substantiates that the introduced algorithm surpasses conventional techniques in terms of performance across varying self-attention head depths, exhibiting enhanced generalization capabilities. Furthermore, an assessment of the algorithm's efficacy is presented across diverse data volumes, revealing a consistent enhancement in performance as the training dataset size augments. The findings of this research offer innovative perspectives and methodologies for tackling feature extraction and classification challenges within extensive high-dimensional datasets, contributing significantly to both theoretical foundations and practical applications.