Artificial Intelligence By Example

accuracy

The accuracy uses the output of the correct prediction to compute the mean of the elements of the tensor in the following code:

with tf.name_scope('accuracy'):accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))tf.summary.scalar('accuracy', accuracy)

tf.reduce.mean calculates the mean in a standard way. For example, the mean of 2 and 4 is 3. Reducing the values of the input gives a good picture of how the model is doing during training without overloading the designer with too many values to analyze, as shown in the following diagram:

Accuracy scalar

The accuracy scalar is sent to accuracy, which can be viewed in the ...