MATLAB: Fundamental Concepts – Neural Networks

Deep Learning Toolboxneural network

If I configure 2 patternnet networks exactly the same – and force them to start with the same input and layer weights, will I always get the same result after training them with the same data ?

Best Answer

Yes, provided the random number generator is initialized to the same state,

(Data division is random as well as the initial weights)

Thank you for formally accepting my answer

Related Solutions

MATLAB: Can we execute neural network in loop

Yes.

I typically loop over

1. Number of hidden nodes (outer)

2. Number of trials for Random initial weights and random data division (inner loop)

Results are stored in matrices of size [ Ntrials numH ]

Searching with my name and Ntrials should bring up examples.

Hope this helps.

Thank you for formally accepting my answer.

Greg

MATLAB: How to generate initial weights for back-propagation neural network (BPNN) using MATLAB R2012a

The reason why the answers are different is because the state of the random number generator changes each time it is called.

Therefore if you want reproducibility, initialize the state of the rng before the first call of configure or train.

Most of my serious designs involve searching over multiple designs for the best combination of random initial weights, random trn/val/tst data divisions and number of hidden nodes.

I have posted scores of examples in the NEWSGROUP and ANSWERS. Search on

 greg rng('default')
 or
 greg rng(0)

Hope this helps