Usage of metrics

A metric is a function that is used to judge the performance of your model. Metric functions are to be supplied in the metrics parameter when a model is compiled.

model.compile(loss='mean_squared_error',
              optimizer='sgd',
              metrics=['mae', 'acc'])

from keras import metrics

model.compile(loss='mean_squared_error',
              optimizer='sgd',
              metrics=[metrics.mae, metrics.categorical_accuracy])

A metric function is similar to an loss function, except that the results from evaluating a metric are not used when training the model.

You can either pass the name of an existing metric, or pass a Theano/TensorFlow symbolic function (see Custom metrics).

Arguments

• y_true: True labels. Theano/TensorFlow tensor.
• y_pred: Predictions. Theano/TensorFlow tensor of the same shape as y_true.

Returns

Single tensor value representing the mean of the output array across all datapoints.

---

Available metrics

binary_accuracy

binary_accuracy(y_true, y_pred)

---

categorical_accuracy

categorical_accuracy(y_true, y_pred)

---

sparse_categorical_accuracy

sparse_categorical_accuracy(y_true, y_pred)

---

top_k_categorical_accuracy

top_k_categorical_accuracy(y_true, y_pred, k=5)

---

sparse_top_k_categorical_accuracy

sparse_top_k_categorical_accuracy(y_true, y_pred, k=5)

---

Custom metrics

Custom metrics can be passed at the compilation step. The function would need to take (y_true, y_pred) as arguments and return a single tensor value.

import keras.backend as K

def mean_pred(y_true, y_pred):
    return K.mean(y_pred)

model.compile(optimizer='rmsprop',
              loss='binary_crossentropy',
              metrics=['accuracy', mean_pred])