2013-2014 Colloquium Series
This Colloquium Series is jointly sponsored by the math and computer science programs. Unless otherwise noted, all talks take place in Jepson Hall, Room 109 at 4 p.m.
Starting at 3:30 p.m. refreshments will be served in the common area outside of Jepson 212 prior to all talks.
More talks will be added as they are scheduled. Please check back.
Students: Becca Funke, Ningxi Wei, John Clikeman, and Gavin McGrew (Faculty Mentor: Dr. James Davis)
Title: Constructions of Bent functions on 8 variables
Abstract: Bent functions are useful in both coding theory and cryptography applications. We will briefly describe why they are important and how to construct them. We then summarize our efforts this summer to find a construction technique that will cover all bent functions on 8 variables.
Students: Natalie Pollard and Jackson Taylor (Faculty Mentor: Dr. Barry Lawson)
Topic: Investigating the "Consistent Programmer Hypothesis"
Abstract: This work involves decompiling binary code into source code (e.g., Java) to determine whether features are preserved in the compilation/decompilation process sufficient to uniquely identify the authors of programs, with implications in identifying authors of malicious code.
Students: Erin Geoghan, Kevin Erb, Richuan Hu (Faculty Mentor: Dr. James Davis)
Title: Difference sets in groups of order 256
Abstract: Difference sets are related to bent functions. We considered the question of which of the 56,092 non isomorphic groups of order 256 contain a difference sets. We summarize our new constructions and we point to future work.
Student: Jocelyn Xue (Faculty mentor: Dr. William Ross)
Title: Smashing Algebras of Toeplitz Matrices
Abstract: Jocelyn will explore algebras of Toeplitz matrices and when these algebras can be multipled together to form all of the matrices
Student: Josh Fagan (Faculty mentor: Dr. Arthur Charlesworth)
Title: Deep Neural Nets Containing Restricted Boltzmann Machines
Abstract: How does the capability of a brain relate to the architecture of that brain? Our project seeks answers to that question, for the current most powerful kind
of artificial neural network (ANN): a so-called Deep Neural Network (DNN). The use of a traditional ANN requires that humans do the work of discoveringfeatures within the data that the ANN can then use to learn to "understand" the data. In contrast, a DNN is supposed to automatically discover featureswithin the data that support the traditional machine learning which a DNN can then use to "understand" the data using such features. Our specific goal isto experimentally investigate a possible relationship
between the number of levels of hierarchies of features within the data and the number of layers ofneurons in a DNN required to successfully "understand" the data. To support our experiments with DNNs, we are developing software to produce potentiallymassive amounts of synthetic data, such that we can experimentally control the number of levels of hierarchies of features within the data.
Student: Georgi Lekov (Faculty mentor: Dr. Kelly Shaw)
Title: Software Optimization: Determining the Best Processor Architecture for an Application
Abstract: Multi-core processors have become increasingly common, but most applications still fail to take advantage of multiple cores. We will present a tool that we created,
intended to help a programmer locate parallelizable sections of code and suggest whether the code would be better parallelized on a set of CPUs or on GPUs.
Student: Nicholas Taylor (Faculty mentor: Dr. Lewis Barnett)
Title: Identifying Bird Species from Digital Images
Abstract: This talk describes techniques for analyzing digital images captured by a motion activated camera to identify birds present in the images. Results will be presented for
identification using the Eigenfaces approach for different preprocessing options.
Maria Nazareth, Senior Manager - Information Systems Management
Adam McCann, Specialist Master - Information Systems Management
Pete Holland, Consultant, UR Class of 2010 – Business Administration (Finance and Economics)
Sarah Gehrke, Consultant, UR Class of 2011 – Political Science; Economics
Danielle Taylor, Analyst, UR Class of 2013 – Chemistry; Leadership Studies
Title: A Data Driven Future - Deloitte Analytics
Abstract: Organizations are seeking ways to translate their data into information, manage performance — and make more effective decisions. This colloquium will provide a background of this field, including specific interactive examples and tools.
The first 45 minutes of the colloquium will introduce the concept, methods, and tools of business analytics, and demonstrate how Deloitte utilizes analytics to provide solutions to our clients’ challenges. Also during this time, practitioners will share how they have used data analytics and tools, such as Excel and Tableau, on their client site and share information about their role with Deloitte, such as our Highly Immersive Visual Environment (HIVE). The final 15 minutes will be reserved for Q&A. Students are encouraged to provide insights and opinions regarding data displays and concepts discussed during the event.
A great opportunity to learn about collaborative research with faculty and fellow students!
Student: Beknazar Abduvaliev (Faculty Mentor: Dr. Michael Kerckhove)
Title: Mathematical Models of Bounded Rationality
Abstract: Models of bounded rationality are intended to demonstrate how decision makers systematically do not make correct inferences and do not choose "rational" actions, due to limitations in processing, receiving or transmitting information and knowledge. I will explore some of the common examples and paradoxes of bounded rationality to show its implications on actions of decision makers.
Student: Nianchen Han (Faculty Mentor: Dr. Joanna Wares)
Title: “Collectivism” or “Individualism”? The Differential Equation Model vs the Agent Based Model
Abstract: The goal of our summer research was to understand the underlying mathematical connection between differential equation (DE) type models and agent based models (ABM). DE type models are easier to analyze than ABMs. However, ABMs may be more appropriate when population sizes are small or when dynamics are not continuous, and are frequently used to verify the validity of DE models under these and other circumstances. In many cases, both model types can be used to describe a population whose size changes stochastically. If they correspond properly, the solution of the DE will be the average behavior of the ABM. We will describe the underlying stochastic process and use two case studies of the simple birth problem and an epidemic control problem to show how the two model types relate.
Student: Uthaipon (Tao) Tantipongpipat (Faculty Mentor: Dr. James Davis)
Title: Ebert's Hat Game with More Than Two Colors
Abstract: The simple version of Ebert's hat game has 2 colors and $2^m-1$ players, and the optimal solution to the game is tied closely to the 1-error-correcting binary Hamming Code. However, there is little research on the solution to the game with more than 2 colors, and the general solution is still an open problem. We will present the optimal solution to the 3-player $k$-color hat game. In addition, we will construct a combinatorial strategy for general numbers of players and colors, and then we will compare the winning probability with Lenstra and Seroussi's linear algebra strategy.
Title: Topology and Dynamical Systems
Abstract: This is a discussion of the role Topology plays in Dynamical Systems intended for a general undergraduate audience. In Dynamical Systems one studies things like a single continuous function f from [0,1] onto [0,1] and sequences like f(a), f(f(a)), f(f(f(a))),. . . . These sequences are called orbits. Exploring the nature of orbits will draw you into the world of fractals and chaos. In Topology we study the structure of the objects generated by Dynamical Systems particularly fractals related to the orbits that do or do not converge, attractors or limiting sets of the orbits, and the entire space filled with the orbits. The orbits live in an infinite dimensional space presenting a special visualization challenge.
Title: A Mathematical Adventure through the Census, Reapportionment, and Re-districting
Abstract: We vote, but how are our votes tallied to give the winner? In 1787, the Constitutional Convention established our rather unusual electoral college which ne-
cessitates an assignment of representatives to the states; how is this allocation done? After giving an introduction to congressional reapportionment and redistricting, she
will explain how mathematics is currently used by the states to prevent and detect gerrymandering. We'll finish with a quick overview of some other ways in which
mathematics and democracy interact.
Bio: Karen Saxe is on sabbatical this year, serving as the AAAS/AMS Congres-sional Science and Technology Policy Fellow. In her normal life, she is Professor
of Mathematics at Macalester College, where she teaches calculus, analysis courses, introductory statistics, and math & politics courses. Her teaching skill has been
recognized with the Mathematical Association of America North Central Section's Distinguished Teaching Award, and with the Macalester College Excellence in Teaching
Award. She is current Editor of the MAA's Anneli Lax New Mathematical Library, and is on the editorial board of the MAA's Math Horizons. Karen has been a resource
in Minnesota on redistricting, consulting with city governments, and recently served on Minnesota Citizens Redistricting Commission, created to draw congres-
sional districts following the 2010 census.
Taylor Applebaum and Andreea Iovan will talk about their experiences this summer as a Software Engineering Intern at Google and Technology Analyst Intern at Goldman Sachs.
Taylor will touch on the experience she has gained from her many interviews at top tech companies, and Andreea will focus on the skills she developed in order to navigate her first industry experience in computer science.
Title: On Life as a Collaborating Statistician at the Mayo ClinicAbstract: The Mayo Clinic is a world premier health care institution filled with people who are passionate about the patient care they provide as well as the research they do. I'll first describe its humble beginnings and its focus on education and research as important factors in providing the highest quality health care possible. Statistics plays a vital role in the research performed at Mayo. I will describe how I got to be a collaborating statistician at Mayo and describe a typical day on the job. I will walk through a project I worked on together with an analytical chemist, and highlight how some basic statistical concepts were used to understand the problem. I will also share some of the things I learned in school that I find most useful today. There will be time at the end for a question and answer session.
Title: Linking genes to diseases: how statistics come into the game
Abstract: What genes are associated with hypertension? Modern biomedical technologies have opened up numerous opportunities for studying the genetic basis of complex human diseases and phenotypes. The availability of ever-increasing biomedical data is calling for cutting edge statistics research. In this talk, we will illustrate the interplay between statistics and biology by exploring statistical hypothesis testing approaches to detecting genes associated with human traits. To address emerging statistical challenges involved in genetic association testing, we will then discuss a new method featuring improved robustness and fast implementation. The method has been successfully applied to identifying genes associated with cholesterol level. This talk will also shed light on what statistical research typically involves and how statistics thrive in multidisciplinary collaborations.
Title: Toward Algebra as a General Problem-Solving Technique: Renaissance Developments from Rafael Bombelli to François Viète
Abstract: By the end of the sixteenth century, algebra had begun not only to take on a whole new look but also to coalesce into a general problem-solving technique, the objective of which was `"to solve every problem. Earlier in the sixteenth century, techniques had been devised to solve cubic and quartic equations, but whereas the proofs treating the cubics continued to rely on geometrical constructions, those of the quartics had had largely to leave the geometrical standard of rigor behind. Enter Bombelli. He was content to leave geometry even farther behind in order to explore complex numbers as actual roots of equations, while his younger contemporary, Viète, continued to preserve vestiges of those earlier geometrical standards in interesting ways. Modern eyes, however, unquestionably recognize Viète's late-sixteenth-century demonstrations as algebraic and not geometric and his approach as one geared more toward general theory-building than toward individual problem-solving. As the work of these mathematicians reflects, the sixteenth century represents a fascinating stage in the evolution of an algebra in the West increasingly distinct from geometry, an algebra on the verge of becoming a new mathematical discipline.
Abstract: Given a collection of linear operators on a Hilbert space, one can ask what other operators can be obtained when operators from this collection are combined though linear combinations and composition. In this talk I will examine this question for Toeplitz operators and Toeplitz matrices.