Problem
I. Split the training set, a validation set, and a test set using stratified sampling to ensure that there are the same number of images per person in each set. Provide your rationale for the split ratio
II. Discuss the rationale behind how your team preprocess the data. Include the mathematical equations used and any dimensionality reduction applied to the instanced and discuss its relevance to the problem at hand. Note that your team will receive more points if you perform data preprocessing that help improve the eventual training process.
III. Select a clustering technique taught in this course and apply it on the training instances. Provide the rationale behind your team's choice of clustering technique and how your team tuned the parameters for the technique implemented.
IV. Clearly discuss - with illustrations - the architecture your team has selected for training and predicting the test instances. Discuss the rationale behind your team's choice of activation functions, loss function, and how you tuned the hyperparameters of the network model.
V. Discuss the results of the trained system.