Matthew L Malloy
My expertise and research includes machine learning, data science and signal processing with applications in internet measurement and digital advertising. I am a scientist at Amazon, and before that, I was the Director of the Machine Learning and Signal Processing master's program in the ECE department at the University of Wisconsin. Before that, I was a Director and Principal Data Scientist at Comscore where I lead a team of data scientists and software engineers. And even before that, I was a post-doc and a PhD student, and before that, I was an engineer at Motorola. My research interests include:
internet measurement and building enormous graphs (see internet device graphs, graphing by example, and crumbling cookies -- which won best paper award at AdKDD 2019)
more scholarly satisfying topics -
learning theory and adaptive sensing (see adaptive sensing and bandits - Lil'UCB)
information theory, optimization, signal processing (estimation via convex relaxations)
I'm very interested in building machine learning algorithms that are gracefully wrong, and correct in aggregate (see ad blockers)
A fun reflection on the 10 billion dollar industry impacted by device graph work