An Improved Integrated Clustering Learning Strategy Based on Three-Stage Affinity Propagation Algorithm with Density Peak Optimization Theory
Table 2
The process of the DPKT-AP algorithm.
Input: similarity matrix S (i, j), cut-off distance dc value, and initial parameter k
Output: final cluster number, division result C= {C1, …, Ck}, and the value of the evaluating indicators
Step 1: select dc value
Step 2: density peak algorithm is used to calculate the local density ρ value and δ value
Step 3: according to the local density ρ value and δ value, the DP algorithm is used to get the initial clustering center point
Step 4: using the k-means algorithm to iterate the data sample and obtaining the several relatively small spherical subgroups, each subgroup has a local density maximum point, which is called the center point of the subgroup
Step 5: run the AP algorithm to go to the third stage of the clustering process, and use the evaluating indicators to evaluate the effectiveness of the algorithm
