Skip to main content

CSAFE Student Researcher Wins Poster Competition at SDSU Data Science Symposium

Ashlan Simpson, a senior in statistics at Iowa State University working at the Center for Statistics and Applications in Forensic Evidence (CSAFE), won first place in the student poster competition at the 2023 South Dakota State University (SDSU) Data Science Symposium.

The symposium was held Feb. 6-7 in Brookings, South Dakota, and hosted by the SDSU Department of Mathematics and Statistics.

Simpson’s poster was titled “Two-Stage Approach for Forensic Handwriting Analysis” and co-authored by Danica Ommen, a CSAFE researcher and assistant professor in statistics at Iowa State.

According to Simpson, her research project aims to identify if two handwriting samples came from the same person or two different people. Simpson and Ommen looked at using the two-stage approach, as it allows for the quantification of errors. Prior research had loosely defined error rates causing difficulties with method assessment and leading to difficulties in justice proceedings.

“This research could get us one step closer to the ability to make a classification of same writer or not with low and quantifiable error rates,” Simpson said. “This would hopefully lead to widespread and effective use in the criminal justice system causing increased accuracy in trial outcomes.”

Federico Veneri, a CSAFE Ph.D. candidate in statistics, also presented at the Data Science Symposium. His oral presentation, “Ensemble of Score Likelihood Ratios for the Common Source Problem,” was co-authored by Ommen.

According to Veneri, machine learning has become a popular alternative to develop comparison scores for forensic evidence; however, the complex dependence structure generated by multiple comparisons has often been overlooked.

Veneri and Ommen’s work proposes a source-aware sampling step to create multiple base learners and ensemble their conclusion into a final ensembled score likelihood ratio (SLR) to enhance traditional SLR. Through simulations using handwriting data, they show that ensemble SLR can outperform the traditional approach.

Ommen gave an oral presentation titled “Statistical Discrimination Methods for Forensic Source Interpretation of Aluminum Powders in Explosives.” It was co-authored by Chris Saunders, a professor of statistics at SDSU, and JoAnn Buscaglia, a research chemist at the Department of Justice and a member of CSAFE’s Research and Technology Transfer Board. The research was supported in part by the FBI Visiting Scientist Program and a grant from the National Institute of Justice.