RIn-Close

Enumerative algorithms for biclustering.

Related works:

• Veroneze, R., & Von Zuben, F. J. (2021). Scalability achievements for enumerative biclustering with online partitioning: case studies involving mixed-attribute datasets. Engineering Applications of Artificial Intelligence.
• Veroneze, R., & Von Zuben, F. J. (2020). New advances in enumerative biclustering algorithms with online partitioning. arXiv preprint arXiv:2003.04726.
• Veroneze, R., & Von Zuben, F. J. (2018). RIn-Close_CVC2: an even more efficient enumerative algorithm for biclustering of numerical datasets. arXiv preprint arXiv:1810.07725.
• Veroneze, R., & Von Zuben, F. J. (2017). Efficient mining of maximal biclusters in mixed-attribute datasets. arXiv preprint arXiv:1710.03289.
• Veroneze, R., Banerjee, A., & Von Zuben, F. J. (2017). Enumerating all maximal biclusters in numerical datasets. Information Sciences, 379, 288-309.
• Veroneze, R. (2016). Enumerating all maximal biclusters in numerical datasets= Enumerando todos os biclusters maximais em conjuntos de dados numéricos. PhD Thesis.
• Veroneze, R., Banerjee, A., & Von Zuben, F. J. (2014). Enumerating all maximal biclusters in numerical datasets. arXiv preprint arXiv:1403.3562.

Compiling

To compile the program, run the file './MakeFile.sh' in the directory containing the source files.

Running

To run the program, type './RInClose' and the arguments:

1 - Dataset's filename;

2 - Algorithm (options: cvcp, cvc, cvcma, chvp, chvpm, chv, opsm);

3 - Minimum number of rows (minRow);

4 - Minimum number of column (minCol);

5 - Epsilon or filename for file with epsilons;

6 - Output filename for the list of biclusters;

7 - Class labels' filename (optional);

8 - Confidence [0,1] (when using class labels);

9 - Ignore biclusters with label x = ? (when using class labels);

Examples are availabe in the folder 'examples'. The example dataset 'dataset' illustrates the input format.

Algorithm options:

cvcp: perfect biclusters with constant values on columns;

cvc: biclusters with constant (similar) values on columns - it uses the same value of epsilon for all columns;

cvcma: biclusters with constant (similar) values on columns - it can use different values of epsilon accross the columns;

chvp: perfect biclusters with coherent values - additive model (shifting biclusters).

chvpm: perfect biclusters with coherent values - multiplicative model (scaling biclusters) / coherent sign changes on rows and columns.

chv: biclusters with coherent values - additive model (shifting biclusters).

opsm: order-preserving submatrices

Note: The problems of mining shifting and scaling biclusters are equivalent. Using an algorithm to mine shifting (scaling) biclusters, we can mine scaling (shifting) biclusters by previously taking the logarithm (exponential) of all entries of the data matrix.