My overall interest is to investigate the development of deep learning methods for improving human health and to explore the limitations of deep learning. Deep learning is an engine of our lab. In one line of research, we are developing explainable deep learning methods for solving a bioinformatics problem known as protein structure prediction. We are currently investigating how deep learning algorithms can be best designed, engineered, and understood for improving existing methods. In parallel, we currently exploring other applications of deep learning including ‘yoga’ and ‘pranayama’ detection and stock price prediction.
Our research is funded by mutiple organizations. The University of Missouri-St. Louis (UMSL) has repeatedly awared me with funds to purchase servers, GPUs, and to pay graduate research assistants. NVIDIA awarded me with a Quardro P6000 GPU device. Google awarded me thousands of dollars worth of Google Cloud Credits to use their GPUs. The U.S. National Science Foundation provided me funds to continue my ‘deep learning for protein distance prediction’ research. In addition to the UM-System’s high performance computing cluster with thousands of computing nodes, we are now fortunate to own a high-end deep learning server with multiple GPUs that we can use for our research. The National Aeronautics and Space Administration (NASA) also provided us with funds to investigate the application of reinforcement learning for detecting gas leakages.