IMath Webinar Series: Mary Elizabeth Segui and Dr. Kenneth Co

You’re all invited to attend the 2nd seminar of our webinar series on Monday, March 27, 2023 at 1:30 PM. Our speakers are Ms. Mary Elizabeth Segui (from the Mathematical Finance and Actuarial Science Research Group) and Dr. Kenneth Co (from Thinking Machines).

Mary Elizabeth Segui (Mathematical Finance and Actuarial Science Group)
Title: Predicting a Browser’s Next Online Behavior Using Markov and Hidden Markov Models with Preference Clustering
Abstract: This research evaluates and compares the accuracy of Markov and hidden Markov models coupled with preference-based clustering in prediction of browser behavior on a vehicle website. Cross validation is performed to determine webpage prediction accuracy of each of the models. Results show that preference clustering based on browsers’ interactions yielded more accurate webpage predictions for both the Markov and hidden Markov models. Furthermore, the Markov model based on interactions with 3 preference clusters yielded the best results among prediction models considered.

Dr. Kenneth Co (Thinking Machines)
Title: Security of Machine Learning
Abstract: The rise in big data and advances in computation have allowed machine learning models to become the algorithm of choice in automating many tasks ranging from image classification, speech recognition, autonomous vehicles, and robotics. However, the unintended risks and vulnerabilities that Machine Learning algorithms bring are not yet fully understood. In this talk, I will discuss a broad range of attacks and defenses against Machine Learning algorithms. This will include topics such as adversarial attacks, data poisoning, model leakage, and model stealing.