Semi-supervised DenPeak Clustering with Pairwise Constraints

Matlab

[cl] = SSDC(X,percent, ML, CL, D)

X: --features, could use load(data);X = zscore(fea); to build features.
percent: --an integer between 0 and 100, sort distances between data points with any other points in an ascending order and select the value in this percent to determined .We use 3% in our experiments as default.
ML: --Must-Link constraints.
CL: --Cannot-Link constraints.
D: --the proposed SSDC with initial temporary clusters, if D is 1, initial temporary clusters. if D is 2, initial temporary clusters.

cl: --clustering results.

[CL, ML] = MakeCons(gnd, N)

gnd: --data labels loaded from dataset.
N: --an integer between 0 and 10, construct (N% * number of data points) pairwise constraints.

You can run one_sample.m as one_sample('Chainlink') to learn how to use this code.

If the code is used in your research, please star our repo and cite our paper as follows:

@article{Ren2018Semi,
  title={Semi-supervised DenPeak Clustering with Pairwise Constraints},
  author={Ren, Yazhou and Hu, Xiaohui and Ke, Shi and Yu, Guoxian and Yao, Dezhong and Xu, Zenglin},
  booktitle={PRICAI 2018: Trends in Artificial Intelligence},
  pages={837--850},
  year={2018},
}