Undergraduates Emily Allen and Benedict Neo present at the National Conference on Undergraduate Research 2023.
The National Conference on Undergraduate Research (NCUR) is dedicated to promoting undergraduate research, scholarship and creative activity in all fields of study by sponsoring an annual conference for students. Unlike meetings of academic professional organizations, this gathering of student scholars welcomes presenters from all institutions of higher learning and from all disciplines. Overall, this conference offers a unique environment for the celebration and promotion of undergraduate student achievement; provides models of exemplary research, scholarship, and creative activity; and offers student career readiness development.
Speaker: Emily Allen, Undergraduate
Title: Statistical Analysis of Kinematic Measurements on Signatures for Forgery Detection
Abstract: As our world becomes more technologically centered, the increased use of digital signatures will heighten the demand for digital signature analysis and forgery recognition. Digital signatures are commonly written on signature pads at various places such as banks and pharmacies. This project aims to develop a statistical approach to differentiate a genuine signature from a forged one. The dataset was obtained using MovAlyzeR, a software that extracts dynamic information from a signature as it is written. The signatures are segmented into strokes, and each stroke's attributes are measured and recorded. The variables collected for each stroke include size (horizontal, vertical, absolute), duration, velocity, jerk, and pressure. This information is used to compute test statistics, which serve as distance measures. These measures are inputted into a random forest classification algorithm that decides if a questioned signature is genuine or forged. This analysis helps identify forgery and can supplement an examiner’s analysis in court cases that involve the crime of forgery.
Speaker: Benedict Neo, Undergraduate
Title: WEPPR: An R Package for the Water Erosion Prediction Project (WEPP) model
Abstract: Soil loss in Iowa is estimated to lose the state $1 billion annually. The water erosion prediction project (WEPP) helps alleviate this issue by estimating soil loss due to water erosion allowing us to implement solutions to counter these effects, ultimately saving money. WEPP is the engine behind the daily erosion project (DEP, dailyerosion.org) which provides daily estimates of soil loss in Iowa, some surrounding regions, and watersheds. Unfortunately running WEPP for all of Iowa takes approximately 6 hours every day as the software takes all the historical data it has on the soil and land data in Iowa and produces a prediction for the current day. We would like to speed up this process by building a Gaussian Process model to emulate WEPP. In order to facilitate this emulation, we are constructing an R package, WEPPR, where users can upload WEPP files to WEPPR and obtain WEPP estimates instantly. This requires the construction of R functions that read in WEPP input files, transform those files to an appropriate format, and visualize results. With the construction of WEPPR and a Gaussian Process model, we will be able to expand DEP from Iowa to the United States and, possibly, worldwide.