I am interested in applying machine learning techniques to various domains (including cybersecurity and finance) and assuring that these algorithms are reliable and safe.
My research focus in graduate school was in modeling the biophysics of protein signalling and conformational change. I've been particularly interested in approaching basic questions of protein biophysics with methods developed in the statistics and machine learning communities, especially nonparametric Bayesian methods.
Counterfactual Explanations for Machine Learning: A Review
Sahil Verma, John Dickerson, and Keegan Hines
NeurpIPS 2020 RSA Workshop (Best Paper Award)
DeepTrax: Embedding Graphs of Financial Transactions
Bayan Bruss, Anish Khazane, Jonathan Rider, Richard Serpe, Antonia Gogoglou, and Keegan Hines
IEEE ICMLA 2019
A Multi-task Network for Localization and Recognition of Text in Images
Reza Sarshogh and Keegan Hines
IEEE ICDAR 2019
Analyzing Single-Molecule Time Series via Nonparameteric Bayesian Inference
Keegan Hines, John Bankston, Rick Aldrich
Biophysical Journal, February 2015
Bayesian Approaches for Modeling Protein Biophysics
My disseration, a real page turner.
Determination of parameter identifiability in nonlinear biophysical models: A Bayesian Approach
Keegan Hines, Tom Middendorf, Rick Aldrich
Journal of General Physiology, March 2014
Inferring subunit stoichiometry from single molecule photobleaching
Journal of General Physiology, June 2013