本文探讨了神经网络的基本概念，包括生物神经元的模拟、常用激活函数（如Sigmoid、Tanh、ReLU等）及其特性，并介绍了神经网络架构的设计原则及如何避免过拟合等问题。
#########
1.Brain analogies
#############
2.Modeling one neruon
Biological motivation and connections
neruro synapses dendrites axopn
activation function sigmoid fuction
#############
3.Coarse model
commonly used activation functions
1.Sigmode function
2.Tanh
3.ReLU(r=Rectified Linear Unit)
4.Leaky Relu
5.Maxout
6.TLDR
Neural Network architectures
Layer-wise organization
1.Neural Networks as neurons in graphs
2.Naming conventions
3.Output layer
4.Sizing neural networks
the neural network can approximate any continuous function.
How to set number of layers and their sizes
problem overfitting :L2 regularization,dropout,input noise
generalization capbility
1.Brain analogies
#############
2.Modeling one neruon
Biological motivation and connections
neruro synapses dendrites axopn
activation function sigmoid fuction
#############
3.Coarse model
commonly used activation functions
1.Sigmode function
2.Tanh
3.ReLU(r=Rectified Linear Unit)
4.Leaky Relu
5.Maxout
6.TLDR
Neural Network architectures
Layer-wise organization
1.Neural Networks as neurons in graphs
2.Naming conventions
3.Output layer
4.Sizing neural networks
the neural network can approximate any continuous function.
How to set number of layers and their sizes
problem overfitting :L2 regularization,dropout,input noise
generalization capbility
扫一扫
437
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
