Talks
Camlis 2019
EMBER Improvements
(slides) (code) (video)
NeurIPS Workshop on Security in Machine Learning 2018
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
(paper) (poster)
DataEngConf NYC 2018
Engineering Lessons Learned by Scientists
(slides) (video)
BSidesSF 2018 & BSidesCharm 2018
EMBER: An Open Source Malware Classifier and Dataset
(code) (slides) (video)
BSidesLV 2017
Data Visualization for Improving Machine Learning Models
(slides) (video)
Data Intelligence 2017
Model Bakeoff
(slides)
SciPy 2015
Examining Malware with Python
(slides) (video)
SciPy 2014
Time Series Analysis for Network Security
(slides) (video with bad audio)
Publications
EMBER: An Open Dataset for Training Static PE Malware
Machine Learning Models
Learning to Evade Static PE Machine Learning Malware Models
via Reinforcement Learning
While at Endgame, I've mostly built and deployed production machine learning systems. But I've also done some research and written some papers.
IGARSS Paper: Ionospheric Stripes in PALSAR data
IGARSS Paper: Simulating SAR through the Ionosphere
I worked on SAR systems for a couple years at a very small consultancy. I found reasons for striping that occurred in SAR imagery collected through the ionosphere.
ApJ Paper: IceCube Search for Muon Neutrinos from GRBs
PhD Thesis: IceCube Search for Muon Neutrinos from GRBs
While getting my PhD in physics, I used the IceCube neutrino detector at the South Pole to search for muon neutrinos coming from giant explosions called gamma-ray bursts. I didn't find any. But I did discover an affinity for building machine learning models.
Press
Wired: AI Can Help Cybersecurity - If it Can Fight Through the Hype
Cyentia Institute Podcast: Episode 9: Phil Roth