-
Notifications
You must be signed in to change notification settings - Fork 0
License
akhtung/curler
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
****************** Parameter Setting of Curler Algorithm ****************** datafile: the dataset file, with no class labels, one line for one data. #cluster: the number of microclusters. #dimension: the dimensionality of the dataset. #data: the number of data points. epsilon_likelihood: probability log likelihood threshold. epsilon_coshare: neighborhood co-sharing level threshold (default 0). microclus file: the microcluser file of #cluster randomely sampled data MaxLoopNum: maximum number of iteration times (2-15 is sufficient). basedim: basedim value of orientation vectors is kept positive (default 0, range [0,#dimension-1]). ****************** Output of Curler Algorithm ****************** output_em.txt: first part: the original data (the d dimension values of each data) second part: the microclusters. (the d dimension values of mean point ) (one eigenvector of d dimensions) (the index of the microcluster) output_expand.txt:the ordered microclusters, one line for one microcluster ( NND nearest neighbour distance) ( nearest microcluster neighbour index (may be not consecutive)) ( d eigen vectors, each of which has d dimension values ) membership.txt: the index of the microcluster to which each data belongs to. ****************** Running Example ****************** iris dataset: Step 1 (dos): clustering Curler iris.txt 150 4 150 0 0 iris_seed150.txt 8 0 Step 2 (matlab): visualization NNCO(150,4,'output_expand.txt'); image dataset: Step 1 (dos): clustering Curler image.txt 500 16 2310 0 0 image_seed500.txt 8 0 Step 2 (matlab): visualization NNCO(500,16,'output_expand.txt'); cubic dataset: Step 1 (dos): clustering Curler cubic.txt 150 2 589 0 0 cubic_seed150.txt 3 0 Step 2 (matlab): visualization NNCO(150,2,'output_expand.txt');
About
No description, website, or topics provided.
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published