For example, here is how I use it in one situation. I exported the model as trainedModel to a file 'trainedModeltexture.mat' using the Export button. This is for the Regression Learner but it will be virtually the same. In fact if you leave the semicolon off the load line, it will give you instructions for how to use it.
% Make predictions based on texture model.
st = load('trainedModeltexture.mat')
% Extract the model from the structure.
trainedModelTexture = st.trainedModel
% Get a table of measurements that we want to make a prediction for.
tTexture = tPredictors(:, 6:end); % Get just columns 6 through the end.
% Do the predictions.
yfitTexture = trainedModelTexture.predictFcn(tTexture);
