File for storing Multi-Layer Perceptron |
File for storing weights and parameters of a MLP.
The file stores the neural net, i.e. its weights and parameters. The file stores the following information:If you want reload the stored MLP, don't edit this file. The suffix of this file is .net.
- Version number and name of the file
- Number of layers, output and input dimension, number of hidden units for each hidden layer and type of activation function
- The matrices of the weights
Example:
The following example stores a Two-Layer MLP with 4 hidden units, three-dimensional input and six outputs:Version: 1 p6.net Number_layers: 2 Dim_output: 6 Act.function: 1 Dim_hidden1: 4 Act.function: 1 Dim_input: 3 Weight matrix output layer : +2.15923085 -2.45975410 -0.88109598 +0.93735627 -0.41550944 +2.08865904 -3.86896036 -0.03048399 -2.78962989 -1.53053199 -1.31088928 +2.07159140 -0.68062739 -3.37259294 +2.79630147 -1.51406900 +2.86396859 +0.58541817 -0.78466628 -2.16585108 -2.52104348 +2.16025013 +0.55498627 -2.79712976 Bias vector: -2.07666933 -0.79373344 +0.16611944 -0.94844541 -1.72233196 -0.75307911 Weight matrix hidden layer : +0.95333027 -2.87205005 -1.18164714 +0.33952172 -0.54502282 +3.39657025 +2.10828810 +0.55708249 -0.89277724 -2.94175113 -0.41389183 -0.96580138 Bias vector: -0.23028497 +0.16500146 +0.38169703 -0.55625551 END
Alphabetic index Hierarchy of classes