Softmax分类器深度学习图像分类方法应用综述

基于深度学习的人工智能图像分类方法研究是当前计算机视觉领域的研究热点。面向深度学习中的Softmax图像分类方法，首先回顾了图像分类技术的发展历程，接着介绍了图像识别技术中的分类器，并解释了Softmax回归函数的分类实现原理。基于Softmax回归分类器的应用，详细阐述了多种图像分类技术，具体包括浅层神经网络、深度置信网络、深度自编码器和卷积神经网络。同时，对比介绍了各种级联模型的具体结构、训练方法、实际应用、分类效果以及优缺点。最后，从Softmax回归分类器、深度学习网络模型和高维数据分类三个方面对基于Softmax回归分类器的深度学习模型在图像分类方面的发展与应用前景进行了展望。

Review of Image Classification Based on Softmax Classifier in Deep Learning

The research on artificial intelligence image classification methods based on deep learning is a hotspot in the computer vision field. Aiming at the Softmax image classification method in deep learning, the development of image classification technology is firstly reviewed in this paper. Then, the classifiers in image recognition technology are introduced and the classification principle of Softmax regression function is given. Furthermore, several image classification techniques based on the application of Softmax regression classifier are described in detail, including shallow neural network, deep belief network, deep autoencoder network and convolutional neural network. Simultaneously, the concrete structure, training method, practical application, classification effect, advantages and disadvantages of above four cascade models are introduced. Finally, the development and application prospects of the deep learning model based on Softmax classifier in image classification are prospected from three aspects: Softmax regression classifiers, deep learning network model and high-dimensional data classification.