When rule-based algorithms do not work? When do we need machine learning (ML) based algorithms? In this post, we will discuss why and when we need to turn to data-driven algorithms instead of using rule based algorithms.
A Convolutional Neural Network trained on the MNIST dataset when presented images from the Devanagari dataset will wrongly classify the Devanagari letter to one of the MNIST classes 0-9, usually with very high confidence. What can we do to fix this?