Conference 2025

Purdue SIAM Student Conference

Conference 2025 Group Photo
SIAM Student Conference 2025 — thank you to all participants and speakers.

Invited Speaker

Prof. Guang Lin

Associate Dean for Research and Innovation · Purdue University

Time: 1:30 – 2:20 PM

Topic: Interpretable, Robust, and Trustworthy AI Systems

This talk introduces several innovative technologies designed to enable the development of interpretable, robust, and trustworthy AI systems. In particular, it showcases AI frameworks that can be reliably deployed for real-time prediction of complex dynamical systems, with applications aimed at enhancing their stability and efficiency. Case studies include COVID-19 pandemic forecasting and personalized prediction in Alzheimer's disease.

About the Speaker: Prof. Guang Lin is the Associate Dean for Research and Innovation and the Director of Data Science Consulting Service. He is the Chair of the Initiative for Data Science and Engineering Applications at the College of Engineering and a Full Professor in the School of Mechanical Engineering and Department of Mathematics at Purdue University. His honors include the NSF CAREER Award, Mid-Career Sigma Xi Award, University Faculty Scholar, and Mathematical Biosciences Institute Early Career Award.

Conference Schedule

9:30 – 9:40 AM

Opening Remarks

9:40 – 10:00 AM
Machine Learning

L1 Regularized Linear Regression with Adversarially Corrupted Data

Analyzes l_1 regularized linear regression under adversarially corrupted training data. Proves deterministic adversarial attacks can be handled with a few samples for support recovery, and studies stochastic adversary models revealing counter-intuitive effects on sample complexity.

10:05 – 10:25 AM
Machine Learning

Privacy-Preserving Relational Learning with Foundation Models

Explores how relational information embedded in graphs can be integrated with foundation models, with a focus on privacy challenges posed by learning from sensitive relational data — particularly when augmenting and fine-tuning pretrained models.

10:30 – 10:50 AM
Machine Learning

Shallow Ritz Method for One-Dimensional Elliptic Problems

Studies the shallow Ritz method for one-dimensional elliptic problems and shows the method dramatically improves the order of approximation for non-smooth problems. A damped block Newton method achieves optimal or nearly optimal order with O(n) cost per iteration.

10:55 – 11:15 AM

Coffee Break

11:15 – 11:35 AM
PDE

ReVAR: Data-Driven Aero-Optic Phase Screen Generation

Introduces ReVAR (Re-Whitened Vector Auto-Regression), an algorithm for data-driven aero-optic phase screen generation that trains on experimental time-series data and synthesizes data matching the statistics of the experimental input — efficiently and with high quality.

11:40 AM – 12:00 PM
Computer Vision

Geometry-Aware Superpixels (GAS) for 3D Scene Understanding

Presents GAS, which extends 2D+time superpixels to efficiently tessellate a 3D volume and dynamically adjusts complexity based on scene content — balancing computational efficiency and high-quality scene understanding.

12:05 – 1:25 PM

Lunch Break

1:30 – 2:20 PM
Invited Speaker

Prof. Guang Lin — Interpretable, Robust, and Trustworthy AI

See invited speaker section above for details.

2:30 – 3:00 PM

Coffee Break

3:00 – 3:20 PM
Mathematical Biology

Parameter Estimation in Agent-Based Models using AABC

Applies Approximate Approximate Bayesian Computation (AABC) to an agent-based model of zebrafish skin patterns and a vertex-based model of meristem development in fern gametophytes, demonstrating AABC's utility for complex biological systems.

3:25 – 3:45 PM
Optimization

Higher-Order Methods for Non-Convex Min-Max Optimization

Examines partially second-order methods for min-max optimization and explores the higher-order smoothness regime by extending higher-order methods to structured non-monotone and higher-order smooth problems.

3:50 – 4:00 PM

Closing Remarks