Diffusion Model Based Research of Clustering Algorithm

Aiming at the various distribute clustering problems in diffusion model for all data points, providing a new clustering algorithm (CDD) based on the change of density. CDD searches the core point using a typical clustering algorithm (DBSCAN) which based on the density, then calculate the direction, speed and acceleration of density diffused which through analyze the diffusion rule of data sample and around the point density, then complete the sample point´s clustering. The experimental results show that: compared with DBSCAN, CDD clustering the diffusion model accurately, and have strong anti-noise-interference ability for the non- diffusion model which make it easier to determine the merits of the parameters.