Ainaz passes Masters Thesis!

Huge congratulations to Ainaz for successfully defending her Master’s Thesis yesterday! (With very minor revisions)

Ainaz’s reserch asked the question of whether we should trust current machine learning based tools for bloodstain pattern analysis (BPA), exploring this interdisciplinary space between computer science and forensic science.

Through her work, she interviewed forensic analysis to establish trust requirements, and carried out a number of technical analyses of state-of-the-art machine learning methods currently proposed for use in BPA, identifying critical robustness weaknesses and an inconsistency of internal model behaviour that varied arbitrarily with factors such as image size – ultimately leading to the conclusion that the reliability and transparency required to be trustworthy in criminal justice settings are currently not present. She published some of this work in the Forensic Science International journal.

Incredible work owning this interdisciplinary space, Ainaz, and demonstrating how interdisciplinary collaboration can lead to real meaningful research results.