Neural Networks: Difference between revisions
Jump to navigation
Jump to search
| Line 12: | Line 12: | ||
The bias unit is optional, but when it is provided, it is always 1. | The bias unit is optional, but when it is provided, it is always 1. | ||
=Multi-Layer Neural Network= | =Multi-Layer Neural Network= | ||
Revision as of 02:27, 4 January 2018
Internal
Individual Neuron
This is a representation of a logistic unit with a sigmoid (logistic) activation function.
The θ vector represents the model's parameters (model's weights). For a multi-layer neural network, the model parameters are collected in matrices named Θ, which will be describe below.
The bias unit is optional, but when it is provided, it is always 1.