John W. Hessler
Mathematician, GIS Scientist, and Lecturer in Washington, D.C.
When not climbing in the Alps, searching through ruins in Central America, or mountain biking through some jungle, I am a Specialist in Computational Geography & Geographic Information Science (GIS) at the Library of Congress in Washington, DC & a lecturer in Quantum Theory and Computing in the Graduate School of Advanced Studies of the Krieger School of the Arts and Sciences at the Johns Hopkins University.
An avid traveler and mountaineer, I am a contributor to Alpinist Magazine, where I write on the history of mountaineering, cartography, high-altitude physiology, climate change and glaciology.
Over the past few years I have lectured on or taught seminars in quantum field theory & computing, evolutionary computation & algorithms, the foundations of computer vision and the mathematics of deep learning. Interested in the interface of computation and neuroscience I have also taught classes in the mapping of the human brain, and the visual cortex & deep networks.
The founder and principal investigator at the Topology Lab for Virtual Geographic Environments, working at the interface of GIS, computer vision and deep learning, we tackle complex theoretical problems in cultural heritage preservation imaging and virtual reality applications in geographical analysis and archaeology.
We are currently working on a project that uses computer vision and Markov Random Fields to reconstruct ancient pottery from fragment assemblages.
My current theoretical research focuses on the mathematics and conceptual foundations of deep learning and on the use of the renormalization group, derived from quantum field theory, and random matrices, to study the complexity of convolutional neural networks. My main line of inquiry concentrates on the topological structure of non-convex loss landscapes and search spaces, along with the mysteries of back propagation.
I am also interested in sparse coding in computer vision research and the application of category theory to problems in GIS algorithm design & quantum computation.
The author of more than one hundred books and articles, including the New York Times best-seller, MAP: Exploring the World, my writing and work has been featured in many national media outlets including the New York Times, Washington Post, Discover Magazine, WIRED, the Atlantic’s CITYLAB, the BBC, CBS News and most recently on NPR’s All Things Considered.
I find being close to supercomputers, higher topos theory, and alpine ice strangely comforting.