Hands-On Transfer Learning with Python

Introduction to transfer learning

Traditionally, learning algorithms are designed to tackle tasks or problems in isolation. Depending upon the requirements of the use case and data at hand, an algorithm is applied to train a model for the given specific task. Traditional machine learning (ML) trains every model in isolation based on the specific domain, data and task as depicted in the following figure:

Traditional machine learning

Transfer learning takes the process of learning one step further and more inline with how humans utilize knowledge across tasks. Thus, transfer learning is a method of reusing a model or knowledge for another related ...