Michael R. Kellman

I am a PhD graduate student at UC Berkeley in the EECS department, where I work on signal processing, computational sensing, compressive sensing, and inverse problems. I am advised by Prof. Michael Lustig and am funded by NSF GRFP. I am an enthusiast of engineering, science, food, and art.

I did my undergrad at Carnegie Mellon University in the ECE Department, where I was advised by Prof. Richard Stern. I have also spent time on Fitbit’s R&D team and at the Labratory of Biomedical Stochastic Physics at NIH working with Prof. Jana Kainerstorfer.

Email  /  CV  /  Github  /  Google Scholar  /  LinkedIn


I’m interested in signal processing, computational sensing, compressed sensing, inverse problems, and optimization. My predominant applications are magnetic resonance imaging and impedance cytometry. Much of my research is about jointly designing physical sensors and signal processing algorithms to solve problems. I have also worked in the area of biologically-inspired speech processing.

Barker-Coded Node-Pore Resistive Pulse Sensing with Built-in Coincidence Correction
Michael Kellman, Francios Rivest, Alina Pechacek, Lydia Sohn, Michael Lustig
ICASSP, 2017

Using communication theory and detection theory, we create a Barker-coded high-throughput microfluidic channel. We jointly design the channel’s system response and cell detection heuristic to efficiently resolve coincidence events by way of an inverse problem formulation.

This work is done in collaboration with the Sohn Lab.

Robust Multi-Pitch Tracking: a trained classifier based approach
Michael Kellman, Nelson Morgan
ICSI Technical Report, 2016

Here we researched the use of correlogram based features to robustly extract multiple speakers’ pitch tracks. These results can be used as input to perform blind speech separation.

Robust Physiologically-Motivated Speech Recognition
Michael Kellman, Richard Stern
CMU Research Symposium, 2015

We research the ability of phyiological-motivated features to perform noise-robust speech recognition. We analyze robustness of mean rate and synchrony features extracted from a auditory-nerve fiber model’s response to speech in the presence of noise.

The Application of pH Sensitive Fluorescent Dye for the Detection of Cancer Cells
Michael Kellman, Moinuddin Hassan, Amir Gandjbakhche
NIH Research Symposium, 2010

We characterized the fluorescence lifetime - pH sensitivity relation for a new pH sensitive fluorescent dye for the application of detecting cancerous skin tissue. The pH differential between diseased and normal tissue, allows us to us this dye to non-invasively track the progression of tumors and cancer treatments. Due to the stochastic scattering effects of tissue and the function response of the system, the direct estimation of fluorescence lifetime is not possible. An inverse problem is formulated and solved to more accurately estimate the fluorescence lifetime of the dye.

Course Projects
Blind Deconvolution: A literature Review and Practitioner’s Guide
Michael Kellman
Numerical Optimization Project, 2017

This goal of this project was to develop an understanding of how a blind deconvolution problem should be formulated and solved. The review gives an overview of several formulations (convex and non-convex), several algorithms to solve them (nuclear norm minimization, interior point method, and alternating minimization), and discusses considerations to memory and computation required to solve large scale blind deconvolution problems.

Remote Monitoring of Heart Rate Using Multispectral Imaging
Michael Kellman, Sophia Zikanova, Bryan Phipps
CMU Signal Processing Capstone, 2015
Poster / Code

This goal of this project was to develop a reliable algorithm to remotely measure heart rate on an Android device. This was accomplished by extracted heart rate from a frontal video of a person’s face, exploiting hue fluctuations in forehead region of face due blood flow.

Carnegie Mellon University ECE 18-220 - Fall 2013 (Lab TA)

© 2017 Michael Kellman. All rights reserved.
Template borrowed from Jon Barron.

Generated by jon-doc