Classification multi-étiquette

Introduction au Deep Learning avec Keras

Miguel Esteban

Data Scientist & Founder

Exemples concrets

¹ https://gombru.github.io/2018/05/23/cross_entropy_loss/

L'architecture

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Instantiate model
model = Sequential()

# Add input and hidden layers
model.add(Dense(2, input_shape=(1,)))

# Add an output layer for the 3 classes and sigmoid activation
model.add(Dense(3, activation='sigmoid'))

Sorties sigmoïdes

# Compile the model with binary crossentropy
model.compile(optimizer='adam', loss='binary_crossentropy')

# Train your model, recall validation_split
model.fit(X_train, y_train,
          epochs=100,
          validation_split=0.2)

Train on 1260 samples, validate on 280 samples
Epoch 1/100
1260/1260 [==============================] - 0s 285us/step 
- loss: 0.7035 - acc: 0.6690 - val_loss: 0.5178 - val_acc: 0.7714
...

Une machine d'irrigation

Passons à la pratique !

Introduction au Deep Learning avec Keras