Self-Attention based Siamese Neural Network recognition Model
Yuxing Liu, Geng Chang, Guofeng Fu, Yingchao Wei, Jie Lan, Jiarui Liu · 2022 34th Chinese Control and Decision Conference (CCDC) · 2022
With the rapid development of machine learning neural network technology, various architectures based on neural network emerge in endlessly. Convolutional Neural Network needs a large number of data to drive to achieve target recognition. In recent years, the Siamese Neural Network model has also been evolving in the development, and has a high recognition effect in the recognition of small samples. Its function is mainly used to measure the similarity of two samples and realize the classification of samples. In order to improve the recognition effect of Siamese Neural Network, this paper will improve the traditional CNN-based Siamese Neural Network, and integrate the self-attention mechanism to achieve better recognition efficiency of the target. After the experimental test model in 4000 data set training, the recognition accuracy can reach 95.16%.