Image classification method for adjusting binary neural network based on weight distribution

• Main Authors: ZHOU XICHUAN, LIU HAIJUN, WANG YUXIAO
• Format: Patent
• Language: Chinese; English
• Published: 19.03.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention relates to an image classification method for adjusting a binary neural network based on weight distribution, and belongs to the technical field of computer vision, and the method comprises the following steps: S1, defining a binary neural network for image classification; s2, configuring extra loss for activating dead weight in the binary neural network, and training a perceived approximate function with a convex derivative to obtain an image classification model; s3, a data set is collected and initialized, and a training set and a test set are constructed; s4, training the image classification model by using a training set, and testing by using a test set; and S5, carrying out image classification by using the trained image classification model. 本发明涉及一种基于权重分布调节二值化神经网络的图像分类方法，属于计算机视觉技术领域，包括以下步骤：S1：定义用于图像分类的二值化神经网络；S2：在所述二值化神经网络中配置激活死权重的额外损失，以及训练感知的、具备"凸"形导数的近似函数，得到图像分类模型；S3：采集数据集并初始化，构建训练集和测试集；S4：利用训练集对所述图像分类模型进行训练，利用测试集进行测试；S5：使用训练完成的图像分类模型进行图像分类。