UFII Symposium

The University of Florida Informatics Institute will host its 4th Annual Symposium on Thursday, March 22^nd, 2018, at the Reitz Union. Students, researchers, faculty, and industry professionals from across the nation are invited to join together to interact, share research, and collaborate. Focusing on the latest trends in informatics, and our Mission of cross-discipline collaborative research, we have invited guest speakers from across the campus and outside the University to present on topics, including election data, text mining, network science, and machine learning.

Location

J. Wayne Reitz Union
2nd Floor, Room 2365

Program

The program will include keynote speakers across a diverse range of scientific fields, with a poster session showcasing UFII Fellows and SEED funded scientists.

8:15-9:00 AM
Breakfast & Check-in

 9:00– 9:30 AM
Dr. George Michailidis
UF Informatics Institute

9:30-9:45 AM
Tyler Richards
DSI Student Group

“Creating Student Led Data Science Curricula”

9:45-10:10 AM
Dr. Emilo Bruna
University of Florida

10:10-10:35 AM
Dr. Ragnhildur I Bjarnadottir
University of Florida

“Text mining for patient safety: What can we learn from nurses’ narratives?”

10:35-11:00 AM
Break

 11:00-12:00 PM
Dr. Vidit Nanda
University of Oxford

“Big Mathematics for Big Data”

Abstract: As data and systems get more complicated, so must the tools with which we analyze them. This talk will describe how algebraic topology has emerged as a powerful new technique for data analysis, and highlight a few of its recent applications.

12:00-1:30 PM
Poster Session & Lunch

1:30-2:30 PM
Dr. Joe Labianca
University of Kentucky

“Social Network Structure, Linguistic Analyses, and Employee Outcomes during a Large Corporate Merger”

Abstract:  We examine language from a corpus of 4 million email communications during a corporate merger in two related studies. We examine the informational advantage inherent in network broker positions in the first study. Brokers gain advantage because of their access to information that is both diverse, covering different knowledge domains, and socially distant, in the sense of coming from different clusters across which knowledge transfer is sticky. We decompose these two elements of the brokerage advantage to understand the extent to which top managers reward employees for employing diverse or socially distant information. We find that brokerage returns are contextually dependent. In the stable, acquiring organization, an employee’s increasing use of diverse information led to salary increases, while the increasing use of socially distant information was being rewarded more in the turbulent, acquired organization. The second study attempts to find a linguistic signature for employees’ perceived job insecurity. We trained a machine learning model to explain employee job insecurity during the highly uncertain merger period. We then applied that model in an entirely different organization in a different industry to successfully predict related outcomes. The imputed job security at the second organization reliably predicts positive events related to job security such as new infusions of cash, and negative outcomes such as involuntary termination. We then deconstruct the model’s “black box” to find which features are most predictive of high or low employee job security. Messages received by employees are as important to predicting job security as the messages the employees author and send themselves, suggesting that multiple underlying mechanisms link job security and these linguistic signatures.

2:30-2:55 PM
Dr. James Hamlin
University of Florida

“Data mining of scientific literature: creation of a database of superconducting material properties”

Abstract: Superconductors are materials that exhibit an extraordinarily useful property: 100% efficient transmission and storage of electrical energy.  The utility superconductors has, to date, been severely limited by the fact that these materials only function at very low temperatures. The discovery of a superconductor that functions at room temperature has been unsuccessfully perused by generations of physicists for over a century.  All superconductors to date have been discovered “by accident” rather than through theoretical prediction. In this talk I will describe work performed by an interdisciplinary collaboration between the Departments of Physics, Materials Science and Engineering, and Computer and Information Science and Engineering.  I will discuss the progress we have made to develop a database of superconductors and materials properties. This database, together with modern high-throughput electronic structure calculations and machine learning, may finally provide the breakthrough needed to design and discover new high temperature superconductors.  Finally, I will compare our approach to that of other projects using machine learning based on purely atomic properties.

2:55-3:20 PM
Dr. Peter Bubenik
University of Florida

“Topological Data Analysis and Hyerspectral Imaging”

Abstract: Topological data analysis uses ideas from the mathematical subject of topology to quantify complicated shapes in images. Hyperspectral imaging takes photographs in which for each pixel instead of just recording color intensities of red, green and blue, instead intensities are recorded for a large number of different wavelengths. I will give an introduction to these two subjects and report on our progress on using topology to detect anomalies in hyperspectral images.

3:20-3:35 PM
Gordon Wilson
Rain Neuromorphics

History

Our previous symposia have featured posters, presenters and attendees from animal science, biology, biochemistry, computer science, ecology, education, electrical engineering, health sciences, mathematics, and political science. We have hosted speakers from industry and other national groups including National Ecology Observatory Network (NEON) and Defense Advanced Research Projects Agency (DARPA). Our goal is to reach out and include a population that is equally diverse. To learn more, visit the 2015, 2016, and 2017 Symposiums.