This seminar talk by Aaradhya Pandey, a doctoral candidate from Princeton University, explores Gaussian certified unlearning in high-dimensional settings. Learn about innovative approaches to machine unlearning that enable removing individual data points' influence from trained models while maintaining privacy and accuracy. Discover how a single Newton step with calibrated Gaussian noise can achieve effective unlearning in high-dimensional regimes where both the number of parameters and samples are large. The presentation contrasts with previous research that suggested multiple Newton steps were necessary, demonstrating that Gaussian certifiability offers advantages in these complex scenarios. The talk takes place on May 22, 2025, at 5:00 PM IST, hosted by the Centre for Networked Intelligence at IISc, with livestream options available through their YouTube channel.