Measure the accuracy of a trained LeNet on the MNIST test data. Set up a neural net and test data:

Now make the measurement:

Take a number of measurements for a trained LeNet on the MNIST test data. Set up a neural net and test data:

Measure the precision and recall:

Measure the macro-averaged precision and micro-averaged recall:

Measure the mean activation of the logits:

Plot the mean activations of the first convolutional layer's filters:

Plot the confusion matrix:

Get the raw data for the confusion matrix:

Plot the average ROC curve:

Plot the ROC curve for class 1:

Measure the error rate of one of the outputs for a multitask trained network for the CIFAR-100 dataset:

NetMeasurements makes use of caching to speed up repeated measurements:

The caching also speeds up similar measurements:

The "IntersectionOverUnion" measurement expects the input and target bounding boxes to be supplied as lists with the form {x₁,y₁,x₂,y₂}, where (x₁,y₁) and (x₂,y₂) are the coordinates describing the bottom-left and top-right corners of a bounding box.

The calculation of intersection over union is equivalent to the following Mathematica function:

Trying to measure an output port being used for the LossFunction can fail:

A simple workaround is to create an additional output port to measure and explicitly label the test data: