Description: Forward neural network (BPNN) 1. First, using a random function of connection weight matrices and bias vectors between each layer of random initialization. 2 in order to use a training sample to train the network, and calculated according to the above formula Δti each sample, t 1, ..., T-1 3. training p samples after (a batch), according to the update equation of W and b are updated. 4. repeat steps 2-3 until the error is less than set threshold or reach the batch number of the set.
To Search:
File list (Check if you may need any files):
BPNN.py