Seminar Schedule (2016 - 2017)

Seminar attendees are invited to gather at 15 minutes prior to the talk to partake of light refreshments and to socialize.

Sep 26, 2016

Ruth Vanderpool

Senior Lecturer, University of Washington, Tacoma

4pm, Thompson 391

 

A Concrete Example of Incompleteness

Abstract: Godel showed that a consistent system expressing elementary arithmetic cannot be complete. Self referential statements, like the liar's paradox, are the first examples often chosen to motivate Godel's theorem. In this talk we will consider a more concrete and arithmetic example by constructing the Goodstein sequence. We will define and compute some terms of the sequence but then see convergence cannot be determined while restricted to Peano's 'finite arithmetic' system.

Oct 3, 2016

Brittany Fasy

Assistant Professor of Computer Science, Montana State University

4pm, Thompson 391

 

Applied Algebraic Topology: Integrating Math, Statistics, Computer Science, and Applications

Abstract: Topology studies the structure of shapes. Topological data analysis (TDA) is the study of the shape of (often large high-dimensional, and noisy) data. Often, in TDA, the data set is transformed into a concise descriptor, such as a persistence diagram or a dendogram, which can then be used to (indirectly) compare or classify data sets. In this talk, we will define a persistence diagram and confidence sets for persistence diagrams. Then, we will discuss how we can use these confidence sets to perform statistical hypothesis testing, and provide a few examples of where we've applied (or are applying) these methods.

 

Oct 7, 2016

(Friday)

Prasad Calyum

Assistant Professor of Computer Science, University of Missouri

4pm, Thompson 391

 

Incident-supporting Visual Cloud Computing with Software-defined Networking

Abstract: In the event of natural or man-made disaster incidents, providing rapid situational awareness through video/image data collected at salient incident scenes is often critical to first responders. Scalable processing of media-rich visual data and the subsequent visualization with high user Quality of Experience (QoE) demands new cloud computing and thin-client desktop delivery approaches. In this talk, we describe the challenges for incident-supporting visual cloud computing and a solution approach for a regional-scale application involving tracking objects in aerial full motion video and large scale wide-area motion imagery. Our solution approach features algorithms for intelligent fog computing at the network-edge coupled with cloud offloading to a public cloud, utilizing software-defined networking (SDN). We will conclude with a discussion of our experimental results collected from GENI cloud testbed that demonstrate how SDN for on-demand compute offload with congestion-avoiding traffic steering can enhance remote user QoE, and also reduces latency, congestion and increases throughput in visual analytics.

Bio: Prasad Calyam is an Assistant Professor in the Department of Computer Science at University of Missouri-Columbia, and a Core Faculty in the University of Missouri Informatics Institute (MUII). Before coming to the university in 2013, he was a Research Director at Ohio Supercomputer Center/OARnet, The Ohio State University. He received his MS and PhD degrees from The Ohio State University in 2002 and 2007, respectively. His research and development areas of interest include: Distributed and Cloud Computing, Computer Networking, Networked-Multimedia Applications, and Cyber Security. He has published over 70 papers in various conference and journal venues. His research sponsors include: NSF, DOE, VMware, Cisco, Raytheon-BBN, Dell, Verizon, IBM, Huawei, Coulter Foundation, Internet2, and others. His basic research and software on multi-domain network measurement and monitoring has been commercialized as ‘Narada Metrics’. He is a Senior Member of IEEE.

Oct 24, 2016

Rainier Aliment

Google

4pm, Thompson 391

 

Finding a Career in Tech

Abstract: I am going to be highlighting my tips for finding an exciting opportunity within the tech community and then plan to do a Q&A with students.

Bio: Rainier graduated from Puget Sound in 2010 with a degree in Molecular and Cellular Biology. He was the sixth member of his family to graduate from Puget Sound. Shortly after graduation, he came back to Puget Sound to work in the Office of Alumni and Parents Relations managing the university's regional alumni programs. In 2012, he joined Google and has held positions focused on staffing operations, technical recruiting and program management.

Nov 7, 2016

Jessica Chan-Ugalde, Sofia Schwartz, and Matthew Bogert

CS Students, University of Puget Sound

4pm, Thompson 391

 

On Diversity and Computing: Insights from This Year's Grace Hopper and TAPIA Conferences

Abstract: In this talk, three computer science students will report back on their experiences at the ACM TAPIA and Grace Hopper 2016 conferences, which celebrate diversity within the field of computing. The talk will discuss current hot topics in diversity and computing (e.g., "biased algorithms" and "cultural edge cases") as well as reflect on the state of diversity within our own Math/CS department and present actionable strategies for empowering students who identify with a marginalized group and creating an intentional community within the department.

Nov 14, 2016

Anderson Nascimento

University of Washington, Tacoma

4pm, Thompson 391

 

Harvesting the Power of Big Data while Preserving Privacy

Abstract: With success stories ranging from online matchmaking to self-driving cars, machine learning (ML) has been one of the most impactful areas of computer science. ML's versatility stems from the wealth of techniques it offers, making ML seem to have a tool for any task that involves building a model from data. And yet, ML makes an implicit overarching assumption that severely limits its applicability to a broad class of critical domains: the data owner is willing to disclose the data to the model builder.

This assumption's significance is immediately apparent in ML applications in security-sensitive industries such as the financial sector and electronic surveillance. In the former, a bank may want to hire an analytics company to mine the data of its customers but, being bound by the customer agreement, cannot simply hand over the data to that company. In the latter, Internet service providers wanting to have a consulting firm do traffic analysis on their logs may be unwilling to disclose details about their customer base in the process. However, perhaps no other business stands to gain as much from lifting the data access assumption or exemplifies its consequences as starkly as healthcare.

ML can revolutionize the healthcare industry by optimizing healthcare spending at all levels from patients to hospitals to governments by addressing these issues. Nonetheless, in practice a major roadblock is that ML for many healthcare tasks (e.g., estimating the risk of hospital readmission) needs data split over many different owners -- healthcare providers, hospitals, and medical insurance companies -- who do not want to or legally cannot share it with outside entities.

In this presentation we will show how recent developments in cryptography, and more particularly in secure multiparty computation can be used to reconcile the benefits of machine learning with privacy for people providing the data behind it.

Jan 30, 2017

Jake Linenthal and Brandon Roberts

Milliman - MedInsight

4pm, Thompson 395

 

What You'll Wish You'd Known about Software Development

Abstract: The primary goal of this talk is to give a snapshot of our professional life and answer questions of students interested in entering the industry. We work on Milliman MedInsight, a platform for doing analysis on all kinds of healthcare data. Some interesting uses of our software include: (1) Evaluating the quality of care provided by physicians, (2) Predicting the probability that someone will be admitted to the hospital, and (3) Finding disparities in care across geographic regions or demographics.

Bios: Brandon graduated from the University of Puget Sound in 2015 with a B.S. in Computer Science. Jake graduated from the University of Puget Sound in 2008 with a B.S. in Math and Computer Science and will receive an M.S. in Computer Science from the University of Washington in 2017.

Mar 6, 2017

Justin Sukkienik

Visiting Professor of Mathematics, University of Puget Sound

4pm, Thompson 395

 

In Search of Transcendental Numbers

Abstract: Classifying real numbers (like, rational versus irrational) can a tricky business. This was Joseph Liouville’s problem in 1844 when he first proved that transcendental numbers existed. He used an approximation method which was later refined by several mathematicians. Finally, in 1955, Klaus Roth found final best possible refinement, which led to discoveries about Diophantine equations. In this talk, we will investigate Liouville’s initial result, examine the existence of transcendental numbers, and connect these results to Diophantine equations.

Bio: Justin graduated from the University of Rochester in 2009. He did a post-doc at the University of Minnesota for three years and taught at Colby College, his alma mater, for three years before joining the faculty at the University of Puget Sound. Justin's research is in algebraic number theory, specifically in Diophantine geometry and height functions.

Apr 10, 2017

Matthew Moreno

Math/CS Student, University of Puget Sound

4pm, Thompson 395

 

Investigating Evolvability in a Genetic Regulatory Network Model

Biological organisms exhibit spectacular adaptation to their environments. However, another marvel of biology lurks behind the adaptive traits that organisms exhibit over the course of their lifespans: it is hypothesized that biological organisms also exhibit adaptation to the evolutionary process itself. That is, biological organisms are thought to possess traits that facilitate evolution. The term evolvability was coined to describe this type of adaptation. The question of evolvability has special practical relevance to computer science researchers engaged in longstanding efforts to harness evolution as an algorithm for automated design. It is hoped that a more nuanced understanding of biological evolution will translate to more powerful digital evolution techniques. This talk will present a theoretical background of evolvability, share recent exciting advances in evolutionary computing, and discuss computational experiments probing the relationship between environmental influence on the phenotype and evolvability.