BEGIN:VCALENDAR
VERSION:2.0
PRODID:ics.py - http://git.io/lLljaA
BEGIN:VEVENT
DESCRIPTION:This talk bridges practical mathematics in defense operations and current research in machine learning. I will outline applied mathematical approaches to manpower modeling in the military\, illustrating how analytical tools shape real-world resource allocation and operational readiness. Shifting to my recent research\, I will present advances in calibrated confidence estimation for neural networks\, focusing on Dirichlet-based classifiers that reduce overconfident errors in large-scale classification tasks. This approach contrasts with the standard use of cross-entropy loss and softmax probabilities\, offering improved reliability for risk-sensitive applications. In such settings\, it is not enough for models to make predictions\, but they must also provide trustworthy measures of confidence\, as poor awareness of prediction quality can lead to costly or unsafe decisions. \n\nAbout the speaker: Courtney Franzen is a PhD student at CU Denver and recently had her paper accepted for an oral presentation at the prestigious IEEE International Conference on Machine Learning and Applications (ICMLA)!\n\nSome undergraduate mathematics or computer science assumed (e.g.\, some calculus\, some programming)\n\n\n\n\n\n\nhttps://today.coloradocollege.edu/events/9172
DTEND:20251010T220000Z
LOCATION:Tutt Science 122 - Lecture Hall (Social Science)
DTSTART:20251010T210000Z
SUMMARY:Mathematics in Mission-Critical Systems: From Manpower Modeling in Defense to Ongoing Efforts to Produce Trustworthy AI
UID:c0cd2868-126b-48ae-aedb-ac3076c7af7d
END:VEVENT
END:VCALENDAR