Finding the accuracy of the model

Accuracy of the model is found by computing mean accuracy over the test set. It is implemented in the following method:

def score(self, test_X, test_Y):
  ...

Here, the parameters are as follows:

• test_X: array_like, shape (n_samples, n_features), Test data
• test_Y: array_like, shape (n_samples, n_features), Test labels
• return float: mean accuracy over the test set

def score(self, test_X, test_Y):
    with self.tf_graph.as_default():
        with tf.Session() as self.tf_session:
            self.tf_saver.restore(self.tf_session, self.model_path)
            feed = {
                self.input_data: test_X,
                self.input_labels: test_Y,
                self.keep_prob: 1
            }
            return self.accuracy.eval(feed)

In the next section, we will look at how DBN implementation can be used on the MNIST dataset.