报告题目：人脸图像的聚类分析

报告人：程明畅 博士 可视化计算与虚拟现实四川省重点实验室

报告时间：5月29日上午9：30-10：00

报告地点：四川师范大学狮子山校区（锦江区静安路5号）第七教学楼C区7楼703

报告摘要：Center-based clustering algorithms such as k-means have strict requirements on data distribution and need to accurately specify the number of clusters. In order to overcome the above limitations, this paper proposes an effective aggregation clustering framework based on the principle component radius (PCR) called PCR-AC. First, k-means++ is used to be the embedding algorithm for initialization. Then, PCR can accurately characterize the core region of clusters to guide cluster merging. On this foundation, a graph based aggregation process can quickly complete cluster identification and improve the local convergence of the embedding algorithm. Experimental results on face datasets verify the superiority of the proposed algorithm and its high tolerance to initialization.

专家简介：程明畅，博士，四川师范大学，毕业于西南财经大学，研究方向为聚类分析、算法设计、机器学习，现已在《Neurocomputing》、《Expert Systems with Applications》等国际期刊发表学术论文若干篇。