Senior Member, IEEE
Senior Lecturer, Department of Computer Science , University of Maryland, College Park.
I received Ph.D. and M.S. degrees in Computer Science in the fields of Computer Vision and Machine Learning from Colorado State University , and B.E. degree in Electronics and Communications from National Institute of Technology, Srinagar, Kashmir. From July 2015 - 2017, I was an Assistant Professor of Computer and Information Science at Harrisburg University, Harrisburg, PA and from 2014 - July, 2015, I was an Assistant Professor of Computer Science & Engineering at the National Institute of Technology, Srinagar. I have also worked as a Software Engineer at Hewlett-Packard, and Wipro. My research interests center on improving and design of systems employing the use of Computer Vision, Machine Learning, and Data Science algorithms.
This project involves development of a Protein Language Model (PLM). The current PLMs are autoregressive models that predict the next token. Although it has been very successful in most of the current applications, there are some, particularly in the development of vaccines where such a model would not be feasible. For example, we may want to produce output around a provided context to not only produce a token after the context but before it as well. Or, a more complex situation where we need to find predictions for context in multiple locations of the given sequence. This is particularly useful in vaccine design. We are proposing a variant of the current transformer-based PLMs that differs from the existing autpregressive models due to its causal masking, positional embeddings and sampling during interference.
The diffusion models produce high fidelity images at the cost of a memory overhead. On the other hand, Generative Flow models (GFMs) use less memory but the quality of the images is poor. Our solution combines invertible diffusion models with GFMs for 3D Medical imaging datasets.
This research involves devlopment of a pair of glasses embedded with RGB-D and LiDAR cameras in combination with a microcontroller to help them navigate around complex surroundings.
Robot navigation in a static environment is a solved problem, however, very often the surroundings are dynamic and the robots would need to interact with a changing environment. We are currently working on such a system in a busy building to help a robot not only navigate its way but interact with its environment along the parth.