Joydeep Medhi was a visiting student from the Indian Institute of Technology, Delhi at Professor Robert Hoehndorf's Bio-Ontology Research Group (BORG) at King Abdullah University of Science and Technology (KAUST). During his three months internship at BORG he worked on "Machine Learning", Classification of causative disease variant(s) with mutations using Machine Learning Techniques. Successfully implemented and explored different Deep Neural Network architectures utilizing large genome sequencing data, interaction networks and ontologies and performed statistical tests for feature selection. He studied Extreme Learning Machines for efficient learning on the imbalanced class distribution. Worked on High-Performance GPU clusters using Tensorflow for implementation and tuning of DNNs.
"Achieved significantly improved ROC AUC 0.967 and Avg. Precision 0.964 on test data for fine-tuned model"