Post-Doctoral Research

§Effectively developed:

o   PON-P2 online prediction tool to decide pathogenic or neutral for variants of human proteins.

o   PON-Diso a machine learning method for disorder prediction of protein at different variants.

o   Specialized predictor for variations of transmembrane portions of proteins.  

o   An iterative method to deduce evolutionary features for protein sequences.

o   Bootstrap of 200 random forests to improve the performance classifier.

§Combined an expert system based on prior knowledge of annotations with random forest classifier for better prediction in PON-P2.

 Ph.D Thesis Project

§Effectively designed an intelligent system for pattern recognition using softcomputing techniques. Contributions are:

o   Noise tolerant, illumination and rotation invariant object recognition using binarized Gabor features and K- Nearest Neighbor classifier

o   Efficient Kannada document segmentation using projection profiles and Gabor filters based segmentation of inflected characters of Kannada script

o   Gabor filters based font type identification for Kannada script

o   Decision tree with modular classifier and 2-stage N-decision trees architecture to recognize Kannada characters

o   Architecture of Kannada OCR System with Braille translation

o   A method to measure the learning capability of a parameter in Artificial Neural Network with application to network freezing and network pruning

o   Optimal Brain Damage a second order measurement for feature vector dimension reduction using Artificial Neural Network