关于非线性激活函数的深度学习分类方法研究

摘要

In this paper, a nonlinear exponential function is proposed as the active coating of the deep neural network. The parameters can be adjusted in the deep neural network to improve the accuracy and other performance. The use of state-of-the-art architectures in the Cifar10/100 database leads to better performance through replacement of activation and adjustment of its parameters. The experimental results show that the proposed nonlinear function can bring about better generalization performance in deep learning.%提出了一种非线性指数函数来作为深度神经网络的激活层,通过调节函数的参数,能够有效提高深层神经网络在分类任务中的测试精度和其他性能.在Cifar10/100数据库中使用最先进的架构进行训练,只通过更换激活函数并调节该函数中的参数能够实现更优的结果.实验结果表明所提出的非线性函数在深度学习中能够带来更好的泛化性能.