[CS] 2016-09-29 Matthew Might
From Mike Donohoe on November 3rd, 2016
Errors in code for software lead to failures both routine and catastrophic -- and to the vulnerabilities at the root of the escalating security crisis. Errors in code for people -- the human genome -- give rise to chronic conditions, devastating rare diseases and, for half of us, cancer. This talk addresses how to end errors in code -- both digital and biological -- through conservatively approximating solutions to the halting problem for the former and through a computational rethink of the practice of molecular or "precision" medicine for the latter.
To evade the halting problem, I will present a broad, universal framework for conservatively approximating the behavior of programs -- Abstracting Abstract Machines (AAM) -- and discuss the success of applying this approach to detecting and eliminating security issues in software.
I will then provide a programmer's introduction and overview of precision medicine; argue that computation has becoming the limiting reagent in saving lives; and explain how a computational re-visioning to the practice of medicine is the key to the diagnosis, discovery and treatment of both rare genetic disorders and cancers.
Speaker: Matt's research in computer science focuses on using programming language principles to improve the safety, security and performance of modern software systems. Matt's research in medicine focuses on diagnosis and discovery of genetic disorders and the use of computing to accelerate and individualize the drug development process. His research is funded by DARPA, NSF, DOE and NIH.
Matt Might is a Strategist in the Executive Office of the President at the White House; a Visiting Associate Professor in Biomedical Informatics at the Harvard Medical School; an Associate Professor in the School of Computing at the University of Utah; an Adjunct Associate Professor of Pharmaceutical Chemistry at the University of Utah; a Co-founder of Pairnomix, LLC, and the President of the NGLY1 Foundation.
Matt tweets from @mattmight and blogs from http://blog.might.net/