Student Paper and Poster Winners

Boston University and Howard University accepted paper proposals from undergraduate and graduate students part of American educational institutions and from any disciplines for consideration for presentation at the 2023 Public Interest Technology University Network Convening

Reviewers selected up to 10 papers from undergrad students and up to 10 papers from graduate students from all submissions. All selected works were peer-reviewed, and could be focused on expanding theory or case studies of real-world applications.


Winning Submissions


Best Undergraduate Paper:

Hannah Ismael, UC Berkeley

Examining Generative Image Models Amidst Privacy Regulations

Abstract: As diffusion models emerge as a new frontier in generative AI, requiring vast image databases as their inputs, the question arises: how should regulators approach policies concerning the collection and utilization of these images? Though generative image models currently interpret the data they scrape as public, regulatory bodies have yet to confirm this as a viable understanding. This paper explores the current public/personal distinction of data as well as the respective legal standards for both categories in both the American and European context. This paper acts as a guide for regulators seeking to understand monopolization and privacy implications of confirming the validity of using open sourced images versus imagining a reality of curated or licensed datasets amidst outrage from artists over a breach of an expectation of collection/use to their artwork. Though arguments have been made regarding using copyright to protect artists, this paper seeks to explore other pathways for regulating generative image models under our current conceptual frameworks of privacy.



Best Undergraduate Poster:

Ujjawal Shah, Howard University

Experimenting with Multimodal AutoML: Detection and Evaluation of Alzheimer’s Disease

Abstract: By 2050, Alzheimer’s Disease(AD) and other dementias could affect 152 million people, increasing social and economic costs for families. To reduce the costs of detection, evaluation, and tracking of AD, this study uses machine learning to predict Mini-Mental State Examination (MMSE) scores, a widely used cognitive impairment screening tool. This work relies on Multimodal learning to analyze the combination of acoustic features and raw translated text to predict MMSE scores as both regression and classification tasks. Applying Principal component analysis to 1733 features, we extract acoustic features. Unlike previous work [1], which heavily relies on manually annotated text to derive features, our approach involves using Google’s Text To Speech API to get audio transcriptions & passing them as input to our multimodal model. The paper presents an improvement in the performance of classification models and a similar performance on the regression tasks. Compared to the baseline [2], our approach improves to an accuracy score of 82% and decreases test set RMSE by 0.28, achieving a final RMSE of 5.56. [1] Aryal SK, Prioleau H, Burge L. AcousticLinguistic Features for Modeling Neurological Task Score in Alzheimer’s. InPACIFIC SYMPOSIUM ON BIOCOMPUTING 2023: Kohala Coast, Hawaii, USA, 3–7 January 2023 2022 (pp. 335-346). [2] Luz S, Haider F, de la Fuente S, Fromm D, MacWhinney B. Detecting cognitive decline using speech only: The adresso challenge. arXiv preprint arXiv:2104.09356. 2021 Mar 23



Best Graduate Student Paper

Mohsin Yousufi

Epistemic Injustice in Technology and Policy Design: Lessons from New York City’s Heat Complaints System

Abstract:  This paper brings attention to epistemic injustice, an issue that has not received much attention in the design of technology and policy. Epistemic injustices occur when individuals are treated unfairly or harmed specifically in relation to their role as knowers or possessors of knowledge. Drawing on the case of making heat complaints in New York City, this paper illustrates how both technological and policy interventions that address epistemic injustice can fail or even exacerbate the situations for certain social groups, and individuals within them. In bringing this case to the workshop, this paper hopes to provide another generative and critical dimension that can be utilized to create better technologies and policies, especially when they deal with diverse and broad range of social groups.