Lex Blog

BlogsAbout
📖 Blogs of machine-learning
All mathematics 19machine-learning 11thoughts 10algorithm 4LHPA 4ai 3computation 2game-theory 2OS 2statistics 2C++ 1cryptography 1frontend 1
非线性与激活函数
线性要求满足可加性与齐次性,但神经网络需要非线性才能摆脱简单的仿射变换。仿射层只能对数据进行拉伸和旋转,而不改变其拓扑结构;激活函数则能实现复杂的形变,使网络能够解开数据流形,有效地学习复杂模式。
mathematics, machine-learning
2026, April, 06
123
RSS Feed
Copyright © 2020-2026 Ethan Zhang. All rights reserved.
AI Policy Sponsor