Ridesharing Car Detection by Transfer Learning’, a paper on transfer Learning based detection of ridesharing cars by Assistant Professor Leye Wang from Peking University, was recently accepted by Elsevier Artificial Intelligence, a top journal in artificial intelligence with a history of 50 years. Particularly, this paper was reported by the news channel of Artificial Intelligence journal. The news channel selects a very small number of papers for reporting each year (one in 2018 and two in 2019). ‘Ridesharing Car Detection by Transfer Learning’ is the first paper reported by the news channel of Artificial Intelligence journal written by Chinese researchers. Co-authors include Professor Daqing Zhang from Peking University, Professor Qiang Yang, Assistant Professor Xiaojuan Ma, and PhD candidate Xu Geng from Hong Kong University of Science and Technology.

Ridesharing has brought great convenience to our life, but the illegal use of ridesharing platforms such as "black car" is still widespread, which reduces user experience, brings security risks, leading to the challenges of the operation of ridesharing platforms. This paper proposes a new ridesharing car detection method called CoTrans, which incorporates the semi-supervised learning (Co-Training) and the deep learning techniques (CNN). CoTrans learns ridesharing driving patterns from other vehicles such as taxis and buses which ware organized by governments. This algorithm can help quickly identify suspected ridesharing vehicles, and then compare with the registered ridesharing vehicles to find the unlicensed ones. This can assist the operation of ridesharing platforms, and improve user experience and security.