Please join us on Friday, November 14th from 12:30 PM to 1:30 PM for the in-person ISP Forum in SENSQ 5317. Lunch and refreshments will be provided, starting at noon. We will be featuring two ISP Student speakers, Zhengbo Zhou and Arun Balajiee Lekshmi Narayanan.

Zhengbo Zhou Presentation Information:

Title: Efficient State Space Modeling of Irregular Longitudinal Mammograms for Cancer Risk Prediction

Abstract: Accurately predicting breast cancer risk from longitudinal screening mammograms is challenging because the data comprise high resolution images acquired at irregular time intervals. Existing approaches either compress spatial detail into per visit vectors or adopt video architectures that are computationally heavy and assume uniform sampling, leaving crucial spatio temporal cues under exploited. We address this gap with a novel state space model designed for irregular longitudinal medical imaging. Evaluated Please join us on Friday, November 14th from 12:30 PM to 1:30 PM for the in-person ISP Forum in SENSQ 5317. Lunch and refreshments will be provided, starting at noon. We will be featuring two ISP Student speakers, Zhengbo Zhou and Arun Balajiee Lekshmi Narayanan.

Bio: Zhengbo is a PhD student in medical image analysis under Prof. Shandong Wu. My research focuses on vision–language models and longitudinal imaging, with an emphasis on breast cancer applications.

Arun Balajiee Lekshmi Narayanan Presentation Information:

Title: Supporting the Study of Program Examples Through Automatically Assessed Self-Explanations and Semi-Structured Dialog

Abstract: Worked Examples are valuable tools for introducing complex problems with solutions to students in programming, physics or mathematics. However, the traditional methods for students to study worked examples are passive. Students could learn more efficiently and engage with the worked example through active learning. An approach to enable active learning in the context of studying programming worked examples is by prompting students to explain code line by line - ``Self-Explanations'' for Programming Worked Examples. Researchers demonstrate that Self-Explanations are a successful learning and knowledge-building approach. However, Self-Explanations work best when a human tutor can provide encouragement and feedback to them. In our work, we present a web-based interface to prompt students' self-explanations to programming worked examples, which they write as they read the code line by line. The system also provides feedback on the students' self-explanations. We conducted a lab study on students' engagement with our web-based code self-explanations interface, Example Student Self-Explanations (WESSE). The study revealed that students were frustrated with the feedback provided by the system, even when they offered correct explanations. We then present a follow-up simulation experiment, ``What-If'', to evaluate and provide feedback to the students' self-explanations using an LLM. This simulation experiment is a potential solution to mitigate the original problem of reducing student frustration, which they expressed in our study.