Problem
Given a face image, we are interested in predicting the identity and age of the person of this face. To this end, we collect a training set with N samples: (Ihy?, (I2, ygj,..., (IN, yN}, where Ln, n = 1, ..., N is the nth image, and 3;." denotes its label. In the training set, we have many different identities and ages. Some images contain the same person but of different ages; some contain different people of the same age.
a) Describe two methods that can transform this face image into a vector.
b) In 1 sentence describe what label 3;", n = 1, ..., N includes.
c) In 3 sentences, explain how you can train an age estimation model from the training set, and how you perform inference given a test image.