Classification iris

MultiLayerNetwork iris classification example.

from miml import datasets
from miml.model_selection import train_test_split
from miml.preprocessing import MinMaxScaler
from miml.deep_learning import Network, LossFunction, Activation, WeightInit
from miml.deep_learning import Sgd
from miml.deep_learning import DenseLayer, OutputLayer

iris = datasets.load_iris()
X = iris.data
y = iris.target

X_train, X_test, y_train, y_test = train_test_split(X,
    y, test_size=0.35, random_state=0)

print(X_train.shape, y_train.shape)
print(X_test.shape, y_test.shape)

scaler = MinMaxScaler(feature_range=(0, 1))
X_train = scaler.fit_transform(X_train)
X_test = scaler.fit_transform(X_test)

model = Network(seed=6, activation=Activation.TANH, weight_init=WeightInit.XAVIER,
    updater=Sgd(learn_rate=0.1), l2=1e-4)
model.add(DenseLayer(nin=4, nout=3))
model.add(DenseLayer(nin=3, nout=3))
model.add(OutputLayer(loss=LossFunction.NEGATIVELOGLIKELIHOOD, nin=3, nout=3,
    activation=Activation.SOFTMAX))
model.compile()

nepochs = 1000
model.fit(X_train, y_train, epochs=nepochs, batchsize=150, print_stride=100)

meval = model.eval(X_test, y_test, batchsize=150)
print(meval)