A Python implementation of Artificial Neural Networks(ANNs) from Scratch.
- This has only two hidden layers.
- This implementation uses the sigmoid activation function.
- Stopping criterion depends on the set threshold.
n_h1
= number of nodes for the first hidden layern_h2
= number of nodes for the second hidden layereta
= learning rate parametermax_epoch
maximum number of iterationsthresh
= threshold, if cost function is less than the threshold, stop the training process.