报告人：白富生 教授 （重庆师范大学）
会议ID：864 975 5986
报告摘要：A novel model for semi-supervised clustering (SSC) problems with pairwise constraints is proposed. The model is formulated as a nonconvex nonsmooth optimization problem. To solve the problem, an auxiliary SSC problem is formulated to generate starting points. An incremental SSC algorithm is then developed. The adoption of the incremental approach allows us to deal with the nonconvexity of the SSC problem by generating good initial points to approximate the solution. The discrete gradient method is applied to solve both the auxiliary SSC and the underlying problems. The performance of the incremental SSC algorithm is evaluated and compared with four benchmarking SSC algorithms on twelve real-world data sets from the UCI Machine Learning Repository. Numerical results show that the presented algorithm outperforms the other four algorithms in identifying compact and well-separated clusters with high constraints satisfaction rate.