Graduate Dissertations

Artificial Intelligence-Based Electricity Demand Forecasting Under Climate Change for Electric Grid Resilience and Adaptation

Artificial Intelligence-Based Electricity Demand Forecasting Under Climate Change Projections for Electric Grid Resilience and Adaptation

Promoting Civic Mindedness and Sense of Belonging to a Glocal Community Through Italian and Italophone Literature With a Migratory Background: A Case Study in the Italian Upper Secondary School Applied Linguistics and Discourse Studies Languages, Cultures, and Literatures Department

"Crafting Climates for Customer Service and Health in the Small Business Craft Beer Industry" Department of Psychological Sciences, Industrial/Organizational Psychology

Title: Towards Robustness in Medical Imaging Classification Abstract: Deep neural networks are increasingly used in medical imaging to assist in detecting and diagnosing a wide range of conditions. Yet current image classifiers remain highly vulnerable to adversarial attacks and natural distribution shifts, which can undermine diagnostic reliability and erode clinicians' trust. They are also susceptible to underconfidence attacks, perturbations that sharply lower model confidence while leaving predicted labels unchanged, creating an additional, often overlooked threat to safe deployment. To address these challenges, this thesis introduces three complementary contributions that improve the robustness and trustworthiness of medical-image classifiers: 1. Robust training with data augmentation (RTDA) - a training algorithm that mixes adversarial training with data augmentation, simultaneously increasing resilience to crafted perturbations and to natural distribution shifts. 2. ConfSmooth - a novel underconfidence attack that reduces a model's confidence while preserving its predicted label, exposing vulnerabilities that bypass existing defense strategies. 3. Underconfidence Adversarial Training (UAT) - an extension of adversarial training that equips classifiers to withstand both underconfidence and misclassification attacks, achieving state-of-the-art robustness.

Doctoral Dissertation Defense of Sara Ailshire Major Advisors: Dr. César Abadía-Barrero Associate Advisors: Dr. Pamela Erickson, Dr. Richard Wilson Title: "Birth Professionals, Human Rights, and Efforts to Improve Childbirth Care In India." Monday, August 4, 2025, 1pm ONLINE To attend virtually via WebEx please use the following link: https://uconn-cmr.webex.com/meet/cea14002

Alessandro Ceteroni, a PhD candidate in Italian Literary and Cultural Studies, Department of Literatures, Cultures, and Languages, will discuss the results of his doctoral research about work in the Decameron.

PhD Biomedical Science Oral Defense Advancing Glycogen Storage Disease Research: Characterization of the GSD-IXβ Mouse Model and Therapeutic Strategies for GSD-Ib.

Dissertation Title: Functional Characterization of USP7 and its Pathogenic Mutants in Neurodevelopmental Disorders. Field of Study: Molecular Biology and Biophysics.

Register for zoom event at:https://us02web.zoom.us/meeting/register/IVfheYU4SUGJid8W4NS5Xw (https://us02web.zoom.us/meeting/register/IVfheYU4SUGJid8W4NS5Xw)