Binary Classifications
Mark Liu · 2023
Binary classification is a machine learning algorithm to classify samples into one of two categories. Examples include whether an email is a spam, whether a credit card transaction is fraudulent, whether a customer will buy a product… In this chapter, readers learn binary classification by classifying images into horses and deer. Readers build a neural network and retrieve the initialized weights in the model. As training progresses, the model weights are adjusted to fit the data. Readers learn how model weights and predictions change over the course of training. In the early stages, if one feeds a picture of a horse into the model, the predicted probability is about 35%. As the training progresses, the probability increases steadily since the model gradually learns from the data. After training, the predicted probability is more than 88%. Readers learn to create an animation to demonstrate how the model weights and the predicted probabilities change in different stages of training.