Hello,
The Neural Network MATLAB Toolbox gives information about the number of hidden layers used but how to find the number of neurons used for training the network? what is the default number of neurons being used? How to change the number of neurons?
Thank you.
The standard network that is used in neural network toolbox is a two-layer feedforward network, with a sigmoid transfer function in the hidden layer and a linear transfer function in the output layer. The default number of hidden neurons is 10 which can changed.
https://www.mathworks.com/help/nnet/gs/classify-patterns-with-a-neural-network.html#f9-26645
Hello Prabhav Borate,
Multilayer backpropagation with NETLAB MATLAB toolbox has functions that allow you to change the number of Neurons used for training,...
Mohamed SOLTANE (Prof. Dr.)
MESRS
Hello,
The standard type used in the artificial neural network in MATLAB, is two layer feed forward network, with 10 neurones with sigmoid transfer function in the hidden layer and linear transfert function in the output layer, but you can change the number and the type of each one, also you can add another hidden layers until you get the optimal network
MatLab has 4 artificial neural network tools according to the objective (at least in 2017a version), so each of them has a different configuration according to its architecture. When you write in the command line or MatLab console: "nftool", "nprtool", "nctool" or "ntstool" you can directly access the wizard of each one and change the number of neurons to use in the layers that you consider. You can also use the wizard to generate a script that designs, trains and executes the neural network; so that it allows to change the functions associated to the different architectures of neural networks, the number of hidden layers and the number of neurons in these. The key to success is to try different combinations until you get the expected results.
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training