A Hybrid Nelder-Mead Method For Biclustering Of Gene Expression Data
M. S. Kavitha, N. Arulanand · International Journal of Technology Enhancements and Emerging Engineering Research · 2016
Biclustering algorithms are used to identify local patterns from gene expression data sets and used to extract biologically relevant information. The fundamental goal of this work is to derive the heuristic approaches to identify the coherent biclusters from gene expression data with minimum MSR (Mean Square Residue) and maximum row variance. Nelder Mead (NM) simplex method is a local search method and very sensitive to the choice of initial points and does not guaranteed to attain the global optimum. The simplex obtained from each iteration continues to shrink and fall into local minima solution. To deal with this problem hybrid optimization approaches namely, Nelder Mead with Levy Flight and Tabu Search with Nelder Mead are proposed and compared. From the analysis, the result shows that NM with levy Flight method performs better to obtain global optima solution when compared and analyzed with NM method and Tabu search with NM.