17. Visualizing Loss

Outputting Training and Validation Loss Metrics

In Keras, the model.fit() and model.fit_generator() methods have a verbose parameter
that tells Keras to output loss metrics as the model trains. The verbose parameter can optionally be set to verbose = 1 or verbose = 2 .

Setting model.fit(verbose = 1) will

  • output a progress bar in the terminal as the model trains.
  • output the loss metric on the training set as the model trains.
  • output the loss on the training and validation sets after each epoch.

With model.fit(verbose = 2) , Keras will only output the loss on the training set and validation set after each epoch.

Model History Object

When calling model.fit() or model.fit_generator() , Keras outputs a history object that contains the training and validation loss for each epoch. Here is an example of how you can use the history object to visualize the loss:

The following code shows how to use the model.fit() history object to produce the visualization. 

from keras.models import Model
import matplotlib.pyplot as plt

history_object = model.fit_generator(train_generator, samples_per_epoch =
    len(train_samples), validation_data = 
    validation_generator,
    nb_val_samples = len(validation_samples), 
    nb_epoch=5, verbose=1)

### print the keys contained in the history object
print(history_object.history.keys())

### plot the training and validation loss for each epoch
plt.plot(history_object.history['loss'])
plt.plot(history_object.history['val_loss'])
plt.title('model mean squared error loss')
plt.ylabel('mean squared error loss')
plt.xlabel('epoch')
plt.legend(['training set', 'validation set'], loc='upper right')
plt.show()

Next Concept