- support any number of layers and neurons in each layer
- use numpy inside to accelerate
Note: all hidden and output neurons are with tanh-type transformation
from nnet import NNet
- prepare data
accept plain list or numpy array.
- define NNet
nn = NNet([n_0, n_1, ..., n_L])
- train NNet
nn.train(X, y)
- get training MLS error
error = nn.score(X, y)
- predict
y_pred = nn.predict(X)