I am currently a PhD candidate in the Data Science Institute at Columbia University with interests in Statistical Machine Learning and Artificial Intelligence. I am co-advised by Prof. Paul Sajda (EE/BME/Physics) and Prof. John Paisley (EE).
Prior to Columbia, I studied Neuroscience, Computer Science, and Biomedical Engineering at Johns Hopkins. I worked in computer vision research in the Vision Lab at Hopkins. I also spent time working on machine learning methods for clinical problems at Oxford Med and Mayo Clinic, and for neural data at Janelia.
Current Research Interests: Variational Bayesian Inference, Deep Generative Models, Probabilistic Computer Vision and ML applications in Healthcare
Amazon AWS AI
Machine Learning Researcher Intern
Jul 2019 - Dec 2020
Engineering Development Group
Jun 2016 - Aug 2016
Howard Hughes Medical Institute
Software Engineer - Scientific Computing
May 2013 - June 2015
Graduate Research Assistant
Jan 2017 - Present
Working on probabilistic machine learning methods for high dimensional time-series data with applications to neural data.
Johns Hopkins University Biomedical Engineering
Research Assistant - Vision Dynamics Lab
Sep 2011 - May 2013
University of Oxford
Research Intern - Oxford Medical School
Dec 2011 - Feb 2012
Summer Research Fellow
May 2011 - Aug 2011
Expanded in-house MRI analysis toolkit; work included developing models of neural atrophy for MRI analysis at Mayo clinic.
High costs, lack of speed, non-intuitive interfaces, and inefficient, fragmented display of patient information have hindered the adoption of the Electronic Health Record (EHR). Critical factors inhibiting adoption of the EMR include the time spent by the health care providers (HCP) in accessing and also documenting patient information during clinical encounters. We describe an emerging visual analytics system dedicated to clinical encounters in emergency room scenarios. It unifies all EMR information fragments into a single interactive visual framework, controlled by voice and touch, in which physicians can conduct diagnostic reasoning tasks in a direct data and information centric manner. We illustrate our system by ways of a typical clinical scenario and point out directions for future research and development.
The common neuro degenerative pathologies underlying dementia are Alzheimer’s disease (AD), Lewy body disease (LBD) and Frontotemporal lobar degeneration (FTLD). Aim of this project was to identify patterns of atrophy unique to each of these diseases using antemortem structural-MRI scans of pathologically-confirmed dementia cases and build an MRI-based differential diagnosis system. We created atrophy maps using structural-MRI and applying them for classification of new incoming patients is labeled Differential-STAND (Differential-diagnosis based on STructural Abnormality in NeuroDegeneration).
Over the last decade the discovery of fluorescent proteins has revolutionized in-vivo imaging in neuroscience. Fluorescent proteins combined with transgenic techniques have allowed researchers to quantify and analyze behavior of various complex cortical circuits across several organisms at Janelia. 2-photon excitation laser scanning microscopy (2PE) has especially aided this endeavor. 2PE has allowed for acquisition of high-resolution, high-sensitivity fluorescence microscopy image’s in intact neural tissue.
ScanImage (3.x, 4.x) is open source customizable software written for 2PE laser scanning microscopes to meet the agile needs of researchers at Janelia and elsewhere.
Optogenetics Acquisition System
ReachTask is software for optogenetic stimulation and data acquisition. It delivers a stable, extensible, software solution for optogenetic stimulation and supports options for integration with 2-photon imaging.
Johns Hopkins University
BA/MS/MS Computational Neuroscience, Biomedical Engineering, Computer Science
2011 - 2016
PhD Machine Learning
2016 - 2021
Copyright Arunesh Mittal 2017