Q-Club Archives
Fall Semester 2013
12/6/2013 Jack Holby and Dan Mulcahey will be speaking about their graph theory project from last Fall. (Last Q club for the semester.)
Steiner Trees: An Introduction and Algorithms
Abstract: We will talk about how Steiner Trees came about (in order to connect points in the plane using the minimum distance possible) and how these particular points can be used in real life applications such as road systems, electric circuits, and other civil engineering projects.
11/15/2013 Robert (Monty) Montgomery will be speaking about his graph theory project from last Fall.
Optimizing Kidney Paired Donation: Optimal and Equitable Matchings in Bipartite Graphs
Abstract: If a donor is not a good match for a kidney transplant recipient, the donor/recipient pair can be combined with other pairs to find a sequence of pairings that is more effective. The group of donor/recipient pairs, with information on how strong a match each donor is to each recipient, forms a weighted bipartite graph. The Hungarian method allows us to find an optimal matching for such a graph.
However, the outcome which is optimal for the group might not be the most equitable for the individual patients involved. We examine several modifications to the Hungarian method which consider a balance between the optimal score for the group and the most uniformly equitable score for the individuals. We examine the strengths and weaknesses of these modifications within the current climate of kidney allocation in the United States. Finally, we expand these findings to other fields where these revised algorithms may also hold particular significance.
11/8/2013 Lara Clemens will be speaking on her summer research experience as part of Research in Industry Projects for Students in Hong Kong.
Robust Low-rank Matrix Factorization
Abstract: Recent advances in signal processing have resulted in new methods for solving the low-rank matrix completion problem (e.g Netflix Prize). Such problems arise in applications involving extremely large data sets that feature redundancy. These data sets can often be modeled by low-rank matrices. As the size of the data grows, solving these problems may become infeasible due to storage requirements. Algorithms which yield factorizations of the low-rank target matrix have been developed and have been proven to be optimal in certain scenarios. However, this method assumes that the matrix to be recovered is not corrupted and satisfies incoherency, or that the observations satisfy the RIP property. In our project, we extend current algorithms to be robust to corruption by sparse noise. We perform numerical tests to support the validity of our methods and apply these algorithms to image colorization and disjoint clique inference.
10/25/2013 Chelsey Legacy will be speaking about her SLU Fellowship (summer research project).
An Interactive Program Using Sums of Rational Functions to Model Correlation Structures
Abstract: Autocorrelation is the correlation between a current observation and a past observation, which is useful to analyze when studying time series data. There are already methods that exist to model autocorrelation structures. However, these structures are not able to model distinct and abnormal features, such as bumps, that occur within an autocorrelation structure. These abnormalities possibly indicate interesting features of the data set that are worth further examination. This project aims to identify the most significant bumps in a series of autocorrelation and model them using sums of rational functions. An interactive R program has been created which allows the user to see a plot of the autocorrelation of a dataset and choose bumps believed to be significant. The program then uses nonlinear regression to calculate the formula used to model the data set with the selected bumps. To illustrate this program, it has been applied to time series data such as data for Disturbance Storm Time Index values, electron flux patterns, and average daily weather temperatures for Anchorage, Alaska.
10/4/2013 Sarah Koallick will tell us about her summer research project integrating Java and R applications.
A Java Graphical User Interface to an R Pareto Front Library
Abstract: A two-stage Pareto front approach is a useful tool when optimizing multiple criteria. In this project, we created a graphical user interface (GUI) that incorporates a collection of R functions written by Dr. Lu Lu and Dr. Christine Anderson-Cook (from the Los Alamos National Laboratory). Their functions find the Pareto front for 2 to 4 criteria and create graphical displays to aid the user in making the best and the most well-informed decision possible. The software we developed presents a GUI for the Pareto front R functions allowing the user, in an understandable way, to input the criteria they want to optimize. The GUI is developed in Java and uses an interface called JRI, the Java/R Interface. JRI allows an instance of R to run inside a Java application.
9/20/2013 Katie Abramski will be speaking about her SLU Fellowship (summer research project).
Improving the Statistical Method for Classifying Geomagnetic Storms
Abstract: Solar storms create disturbances in the Earth’s magnetic field that can damage satellites and cause power grids to fail. In an effort to better understand solar storms, we explore ways to improve the current method for classifying events as storms based on the Disturbance Storm Time (Dst) index. Several methods currently exist for classifying events as storms based on their Dst observation, such as -100nT or -65nT. We investigate how these classification methods were created, and search for a more statistically justified way to define storms. We explore different methods of classifying events as storms based on Dst.
The SLU Fellows program gives students the excellent opportunity to do research and collaborate with others. In addition to presenting my research project, I will talk about the SLU Fellows application process as well as some of the benefits of the program.
9/6/2013 Juan Chang will be speaking about his summer research Project.
Predicting Owner Tendencies in Fantasy Football Drafts
Abstract: Fantasy Football is a popular interactive competition in which participants manage their own virtual National Football League (NFL) teams. This virtual pastime is based on real NFL game statistics. In order to study trends within the Fantasy Football draft, this project uses logistic regression analysis and Classification and Regression Trees (CART). I expect to develop an effective and efficient model that will predict the probability that a position will be chosen given any round and pick in the draft. Ultimately, fantasy football owners may be able to use this model to choose the most successful team in their prospective league.
Spring Semester 2013
5/3/2013 Luke Horton will speak. (Last Q club for the semester.)
Game Development for iPhone
Abstract: The purpose of my senior year experience was to achieve a better understanding of iOS development process, and to publish to the App Store a purchase-worthy mobile game. I began by learning the many tools required for the development process, including the iOS SDK, a proprietary IDE (Xcode), and an open-source game engine (Cocos2d). With a semi-solid understanding of the major concepts driving iOS devices, I started to design and implement a game. Throughout development I met many hurdles and climbed the steep learning curve of app development for iOS. The project successfully concluded with submission to the App Store.
4/12/2013 Luke Reed will be speaking.
Compartmental Models: An Epidemiological Application of Ordinary Differential Equations
Abstract: Ordinary Differential Equations can be used to create epidemiological models to study the spread of infections. The traditional SIR model splits a population into three separate categories: Susceptible, Infected, and Removed. This model provides a basic understanding of the progress of an infection, however, it overlooks important details. By splitting these categories into smaller subgroups, it becomes possible to understand the various stages of the infection and study it within the contexts of populations with varying interaction and transmission rates.
3/29/2013 Spencer Timerman will speak on his research conducted with Dr. Sam Vandervelde.
Divisor Graphs for Egyptian FractionsDescription
Abstract: Exploring simple graphical representations of sums of Egyptian Fractions and obtainable corresponding sets of graphs and fractions.
2/1/2013 Shelley Kandola will be speaking.
She will be presenting her research with Sam Vandervelde on decomposing the real line (similar to Banach Tarski paradox).
2/1/2013 Dr. Lisa Torrey, Assistant Professor of Computer Science
Artificial Intelligence: Success Stories
Abstract: In science fiction, artificial intelligence (AI) is practically limitless. Space ships communicate with their pilots, robots look and move just like people, and computer programs wrestle with ethical and emotional dilemmas. We're not there yet, but research in AI has produced some exciting and useful things. At this week's Q-club, you'll learn about some of the success stories, both old and new.
Fall Semester 2012
12/7/2012 Dr. Heng Yin of the the Department of Electrical Engineering and Computer Science at Syracuse University will present on Android Malware Analysis.
Abstract: The prevalence of mobile platforms, the large market share of Android, plus the openness of the Android Market makes it a hot target for malware attacks. Once a malware sample has been identified, it is critical to quickly reveal its malicious intent and inner workings. In this talk I will present DroidScope, an Android analysis platform that continues the tradition of virtualization-based malware analysis. Unlike current desktop malware analysis platforms, DroidScope reconstructs both the OS level and Java-level semantics simultaneously and seamlessly. To facilitate custom analysis, DroidScope exports three tiered APIs that mirror the three levels of an Android device: hardware, OS and Dalvik Virtual Machine. On top of DroidScope, we further developed several analysis tools to collect detailed native and Dalvik instruction traces, profile API-level activity, and track information leakage through both the Java and native components using taint analysis. These tools have proven to be effective in analyzing real world malware samples and incur reasonably low performance overheads.
11/9/12 Torrey Hayden will be speaking about her summer REU.
Modeling X-ray Studies of Catalytically Active Surface Sites on Nanocrystals
Abstract: Because x-rays have wavelengths around the length scale of an atom, x-ray diffraction is a useful tool determining distributions of atoms within a material. This can be used in a variety of areas, such as chemistry, materials science, physics and biology. It has been shown that changes in structure of a monolayer of tungsten oxide (WO3) on a bulk hematite (Fe2O3) material are discernible through the use of x-ray techniques. We created a computer model of x-ray kinematic diffraction that determines the intensity of diffraction off a nanocrystal from an incident x-ray beam. Using this tool, we simulated a similar experiment by determining the intensity readings a detector would pick up when x-rays are diffracted off of a hematite nanocrystal with tungsten oxide structures a face. When this crystal was used in the model, the diffraction pattern shifted depending on whether the tungsten oxide structures were in their reduced or oxidized states. Hence, we can predict that the changes in the structure of tungsten oxide during a redox reaction should be detectable.
11/2/12 Brian Thomas will be speaking about his summer REU.
Washboard: An Effective Anaerobic Exercise Game
Abstract: To date, researchers have focused on the intensity of exercise games (exergames) using aerobic workouts. It is important to balance workout programs with anaerobic exercises. Unfortunately, no research has investigated the effectiveness of exergames focusing on anaerobic workouts. Therefore, we have created Washboard. The purpose of Washboard is to perform sit-ups in order to pop balloons. Although our study is currently in progress, preliminary results suggest players are getting a hard to heavy workout, while perceiving an easy workout.
10/19/12 Michael Schuckers, Associate Professor of Statistics
Dr. Schuckers will speak about different summer research opportunities for students and the upcoming Senior Seminar for the Spring semester
10/5/12 Joey Kelley
"Evaluating Ridge Regression with Application to Fantasy Football Data"
Abstract: Ridge regression is a generalization of ordinary least squares regression that is sometimes useful for prediction. It is commonly used in cases of highly correlated (i.e. nearly linearly dependent) predictors. As part of a SLU Fellowship, we studied several estimators for ridge regression parameters. We evaluated their performance compared to ordinary least squares regression using simulated data. Using these results, we applied ridge regression to fantasy football data in order to make better player ratings.
9/21/12 Kevin Angstadt will be speaking about his SLU Fellow.
"Developing Interactive Web Tools for Statistics Students"
Abstract: Advances in computing and computer technology have had a profound effect on statistics and statistics education. These advances have made simulation methods a practical pedagogical tool for teaching basic concepts of statistical inference. For my SLU fellowship project, I worked alongside Professors Ed Harcourt, Patti Frazer Lock, and Robin Lock to develop web-based statistics programs, collectively known as StatKey.
9/7/12 Kerrin Ehrensbeck will be speaking about her SLU Fellow.
"Improving Plutonium Management at the Los Alamos National Laboratory"
Abstract: The Los Alamos National Laboratory (LANL) located in Los Alamos, New Mexico controls and manages most of the remaining nuclear material in the United States. An extremely important job that requires the use of highly accurate measuring devices called calorimeter. A calorimeter is a type of heated water bath in which the heat deficit of the plutonium is measured. Unfortunately the plutonium is leaving behind a heat residual which, in turn correlates future plutonium to past measured items. During my SLU fellowship this summer, I explored possible solution for fixing this correlated data through the use of Bayesian Statistics.