Recently, IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), a top international journal in the field of artificial intelligence, published the latest research achievement of Professor Duan Jiang and his team, namely, “Supervised Anomaly Detection via Conditional Generative Adversarial Network and Ensemble Active Learning”. The four authors of the paper are all team members in the China Blockchain Research Center, among whom Associate Professor Chen Zhi is the first author, and Professor Duan Jiang is the only corresponding author.
The study focuses on the critical task of “outliers detection” in the field of data mining. To overcome the challenge of obtaining massive amounts of annotated data required for training and tuning the DEPTH model, the team creatively proposed a method of generating and using “fake data” on a large scale. Compared with the existing 13 most advanced methods, the new method can improve the detection accuracy by 25% to 52% with only 5% of the training data annotated. Such significant accuracy improvement can be of great help to the efficient completion of fraud detection, outliers detection, and other data mining tasks.
TPAMI means the IEEE Transactions on Pattern Analysis and Machine Intelligence. China Computer Federation. Chinese Association of Automation, and other institutions position it as a top journal in the world, encouraging Chinese scholars to make breakthroughs. With the impact factor of 24.314, TPAMI ranked first among computer engineering, electronic engineering and artificial intelligence related journals in 2020.
This is not only the first research achievement published in TPAMI by Southwestern University of Finance and Economics as the first institution, but also another landmark achievement made by the Team of the China Blockchain Research Center of SWUFE, which has always been focusing on the cutting-edge theoretical research of digital technologies such as blockchain and artificial intelligence as well as the applied research greatly needed in China. The Center has been strongly supporting the development of information science related disciplines in SWUFE by a series of achievements made for industries, universities, research institutions and users since its establishment.