Abstract

Transfer learning (TL) is a subfield in machine learning that solves a new learning task by applying stored knowledge to an existing learning task. It has a higher efficiency than training a model from scratch because it normally requires less training data and training time, while achieving relatively good accuracy and reliability. This chapter introduces the definition, motivation, types, and procedure of TL, followed by four case studies in intelligent transportation systems, where the first and second cases are about smart parking problems (vehicle detection and parking pattern prediction), and the third and fourth case studies address data scarcity issues at nighttime and in the real-world by using labeled data from daytime and also simulators.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call