Title :
Enhanced prediction of protein cellular localization sites with genetic algorithm optimal kernel projection analysis
Author :
Isaacs, Jason C. ; Foo, Simon ; Meyer-Baese, Anke
Author_Institution :
Florida State Univ., Tallahassee
Abstract :
The localization of proteins can help us to better understand their functions. Currently a number of localization machine learning algorithms have been employed on this problem, including SVM and K-NN. However, in terms of performance there is little success. In this paper, we apply a genetic algorithm to optimize a kernel component analysis solution, an algorithm that wc believe will have better performance than standalone Kernel PCA. We will experiment using the protein location data from the Horton and Nakai yeast and E.Coli databases. We will then compare the performance of our optimized system with previous methods. Results show that GA enhanced kernel component analysis can improve classification.
Keywords :
biology computing; genetic algorithms; learning (artificial intelligence); e.coli databases; genetic algorithm optimal kernel projection analysis; machine learning algorithms; protein cellular localization sites; Algorithm design and analysis; Databases; Fungi; Genetic algorithms; Kernel; Machine learning algorithms; Performance analysis; Principal component analysis; Proteins; Support vector machines; Genetic Algorithms; Kernel Methods; Manifold Learning; Projection Analysis; classification.;
Conference_Titel :
Evolutionary Computation, 2007. CEC 2007. IEEE Congress on
Conference_Location :
Singapore
Print_ISBN :
978-1-4244-1339-3
Electronic_ISBN :
978-1-4244-1340-9
DOI :
10.1109/CEC.2007.4424639