Bayes Optimal Classifier

Photo by Owen Beard on Unsplash

Machine Learning from First Principles: Blog Post 5

In the last blog we learnt what realizability and sample complexity are. Reiterating it, realizability assumption states that the optimal hypothesis lies in the finite hypothesis class. This is a strong assumption as the universe would not be kind enough to lend us the correct answer from the set we choose to find it from. This assumption is just for understanding and starting simple…