Chongqing University of Posts and Telecommunications
Search
中文 | 한글
News&Events
CQUPT's Research Result Is First Published in TPAMI---the Top Authoritative Journal in the Field of Artificial Intelligence
Time: 2020-07-16   

 CQUPT News (Provided by Division of Science and Technology and edited by Ao Yongchun) Recently, Xia Shuyin, Wang Guoyin and other researchers of Chongqing Key Laboratory of Computational Intelligence of our university published their latest research result “A Fast Adaptive K-means with No Bounds”, which was published online on IEEE Transactions on Pattern Analysis and Machine Intelligence, the top international academic journal in the field of artificial intelligence. This is the first time that our university had published a paper in this authoritative journal, marking that our university had made great progress in basic scientific research in the field of big data intelligence.

 IEEE Transactions on Pattern Analysis and Machine Intelligence is one of the top international journals in the fields of artificial intelligence, pattern recognition, image processing and computer vision. It is a Class-A Journal of China Computer Society (CCF) and JCR No.1 District of Chinese Academy of Sciences. It has authoritative influence in the field of computer science and artificial intelligence, especially the frontier innovation achievements in the field of pattern analysis and machine intelligence. Its current impact factor is 17.861.

A new accurate k-means algorithm is proposed in this paper. K-means is one of the most commonly used fast basic algorithms in artificial intelligence. It is widely used in clustering, data pre-analysis and other machine learning algorithms. On the basis of the multi granularity granular computing theory (Xia S, Liu Y, Ding X, Wang G, Yu H, Luo Y. granular ball computing classifiers for efficient, scalable and robust learning [J]. Information Sciences, 2019, 483:136-152.), which is proposed by Xia Shuyin, Wang Guoyin and Yu Hong, this paper uses hypersphere to divide the metric space to have obtained a more accurate neighborhood relationship. The algorithm does not need additional parameters, and eliminates the upper and lower bounds of single sample in most of the existing excellent acceleration algorithms, and its distance calculation times are less than those of existing similar algorithms.

 This research is jointly completed by experts and young postgraduates—Xia Shuyin, Peng Daowan (master), Wang Guoyin, and Chen Zizhong, along with Meng Deyu of Xi’an Jiaotong University, Zhang Changqing of Tianjin University, Elizabeth Grime (doctoral candidate) of University of California, and Wei Wei of Xi’an University of Technology.

 

This is a basic diagram of near neighbor cluster.