After concluding my PhD program, I worked @KurtLab, University of Washington, to improve tumor identification from 3D brain MRIs using Generative Modeling techniques. I am currently a Postdoctoral Scholar at Computational and Integrative Pathology Group, Northwestern University. I work on application of Active Learning, and Foundation Models for several classification and segmentation problems in the histopathology domain with extremely high-resolution images (~billion pixels).
I completed my PhD from the bionics research lab @ROMBOLABS, University of Washington.
My PhD research was about using Machine Learning and Wearables to analyze, predict and generate human movement in real world environments, with the goal of making assistive devices more expressive. Here is my PhD work in a nutshell. Here is Everything (Almost) You Always Wanted to Know About Lower Limbs* (*But Were Afraid To Ask)
Before starting my PhD, I completed my bachelors in Mechanical Engineering from IIT Bombay.
Towards the end of my bachelors, I developed a keen interest in robotic hands and soft robotics. This along with my interest in mechanical design led to my first humble attempt at research: Design of a Soft Anthropomorphic Hand with Active Stiffness control, before coming for my PhD.
Our environment shapes how we move through space. We have to alter our gait to avoid obstacles and move towards the intended goal. Humans use vision to plan their movement through space. So, can egocentric vision help predict their gait forward in time?
Human bodies are capable of generating a diverse set of movements. We have some intuition about which activites are more complex than the others, for example walking forward in a straight line might be considered less complex than navigative obstacles. Then, there are mathematical notions of complexity for example dimensionality, variability etc. In this work, we explore whether different measures of complexity agree with our intuition.
Everyone has a unique style with which they move. Then why should the assistive devices be generic?
Digital pathology has allowed improved data collection and labeling processes, as well as better data sharing. Consequently, there is a greater scope of using machine learning to build better tools for pathological diagnoses, and analyse digital pathology datasets. The challenges however include more fat-tailed datasets compared to natural domains, higher computational requirements due to large whole slide images, requirement of domain expertise for data labeling, as well as interobserver variability in data labeling. There is also a greater emphasis on developing human-in-loop solutions as compared to fully automated systems.
Medical imaging is crucial for early detection and prevention of fatal diseases. The datasets involved are hard to collect, and much smaller in quantity compared to most computer vision tasks. Can generative modeling techniques alleviate these challenges?
Representation Learning, Recurrent Neural Networks, Convolutional Neural Networks, Autoencoders, Generative Adversarial Networks, Denoising Diffusion Probabilistic Models, Decision Trees, Style Transfer, Spectral Clustering, Semi-Supervised Learning
Wearable Motion Capture, Gait Analysis, Biomechanics, Prosthetic Control.
