Hilldale Lecture's Theory Seminar | UW-Madison Computer Sciences Dept.

Three Of My Favorite Open Problems

I will discuss some open problems that I have worked on forever. I think these problems should be solvable but progress is very limited. The first is essentially the halting problem for linear automata. It has been called a shame that we cannot prove it is either decidable or not. I will report on some partial results. The other two problems I will leave as a surprise, but they are both related to our P vs NP question.

Empowering systems design with end-to-end specialization

Can we speed up a CPU-bound server application that does complex in-memory processing under tight latency budget, by moving most of its in-memory state to Flash storage / SSD (which is ~500x slower than RAM) ? Can we achieve crash-tolerance via online process-pair replication and consistent virtual memory snapshotting in a CPU-bound system, while incurring negligible common-case performance hit? Can we speed up petabyte-scale “big data” query processing in a cluster by orders of magnitude while simultaneously reducing cost?

Software Developer

Bonfire is a fast growing, Madison-based, startup with two UW graduates a
part of our co-founding team. We have an opportunity that we are excited to
share with the UW Computer Science community. We are seeking an
undergraduate or graduate student or group of students who are proficient in
web/mobile application development. We are seeking students with
experience in JavaScript, Ruby on Rails, and/or database development. The
project can count as credit or be a freelance project.

CPCP Privacy/Fairness Seminar: The Bounty of the Commons

This is an exciting time in biomedical data science. It is now possible to collect substantial information about individuals and their encounters with health care. Our ultimate goal is to integrate this data, along with data and findings from those engaged in basic science, to identify new opportunities to improve health. Broad data sharing will further our progress towards this goal. However, data sharing poses both cultural and technological challenges.

Building Argo: An Autonomous, Adaptive Internet Routing Overlay

Abstract: How do you get bits across the Internet 35% faster? At Cloudflare, a small team of engineers in central Illinois built Argo, an autonomous, adaptive routing layer on top of the Internet. Argo continuously measures Internet performance across Cloudflare’s network and intelligently identifies routing paths that are faster than vanilla BGP Internet routing. We solved many technology challenges building Argo — we built out a distributed network proxy capable of scaling to 10% of global web requests and analyzing event streams of hundreds of thousands of events per second.

Pages

Subscribe to UW-Madison Computer Sciences Department RSS