I'm Rich Harang, a Principal Data Scientist/Machine Learning Researcher at Invincea, Inc., where I focus on applying machine learning to problems in host-based security. My main projects include applying deep learning to detecting and understanding malicious code samples in a variety of formats, but Other ongoing work includes techniques for attribution and post-incident analysis. I also have a hobby interest in cryptography as well as security applications of computational complexity.
Prior to working for Invincea, I spent five years working at the U.S. Army Research Laboratory -- two years as a contractor with ICF International, and three years as a federal employee -- on basic and applied research projects at the intersection of machine learning, network security, and computational complexity. Notable work includes authorship attribution for both source code and compiled binaries in collaboration with researchers at Drexel and Princeton, and lightweight methods for intrusion detection in resource constrained devices.
Before working for ARL, I spent a year as a postdoctoral scholar in the Computational Science and Engineering Research Group at the University of California, Santa Barbara, under the direction of Dr. Linda Petzold. While there, I studied the role that stochasticity plays in the behavior of circadian oscillators and networked stochastic processes. I spent some time looking into ways that we can infer topology from observations on the nodes of a network, and exploring how this network structure can allow systems to exploit stochastic behavior to their advantage rather than simply manage it away.
My dissertation, "Wavelet Analysis of Stochastic Circadian Oscillators," focused on using continuous wavelets to infer properties of oscillators driven by solutions to stochastic differential equations. I used the results to analyze bioluminescence data collected from the suprachiasmatic nucleus of rodents, including topological connectivity, classification of neurons, and using stochastic differential equations as a model of the neurons.
For fun, I do a bit of tinkering with computational statistics -- particularly Bayesian methods -- and try to apply them to whatever interests me at the time, very often political science. If you're interested, you can read about it at my personal blog (linked over on the left). It's a bit neglected these days, now that I'm gainfully employed, but I do manage to get something posted every once in a while.
I love talking about my work, and I'm always interested in exploring potential collaborations, so if you've got questions, feel free to contact me.