Rohith Teja M

- Focus on methods to quantify regional budgets of anthropogenic CO2 emissions using ground-based near real-time activity data on energy and mobility. - Carbon Monitor: https://carbonmonitor.org

- Validated and enhanced data within a knowledge graph in the Internet of Things domain using state of the art algorithms for deep graph learning and graph embeddings for knowledge graphs - Achieved excellent performance for embeddings with 98% ROC-AUC and > 80% average precision scores in prediction of missing links in the graph by using an end-to-end machine learning pipeline

- Generated dynamic graph embeddings (numerical representation of proteins) using deep Autoencoders and predicted missing protein interactions in the network with an accuracy score of 80% using SVM - Identified important protein clusters (ranging from 1 to 24 clusters depending on the cell type) from dynamic graph embeddings using K-NN and clustering algorithms

- Machine learning, Deep Learning, Data Analysis, Computer Vision, Big Data & Project Management - Course website: https://mldm.univ-st-etienne.fr/course_organisation.php