Seminar Organizer: Kumar Abhinav
Previous Series Lists [Series 1, Series 2, Series 3, Series 4, Series 5, Series 6, Series 7]
Series 7
Upcoming:
On 19th of August 2025 at 14:00 hrs ICT
Title: Beyond the Bulk: A RMT-based Approach to High-Dimensional Data Reduction
Abstract: Traditional PCA often falls short when dealing with high-dimensional noisy datasets, as it relies on variance maximization, only capturing components with the highest variance. It tends to often confuse high-variance noise for meaningful signal and ignores low-variance patterns that might carry important structural information. To address this issue, we introduce a new method based on Random Matrix Theory (RMT), using the Marchenko-Pastur distribution to statistically separate signal from noise. Unlike PCA, this approach avoids reliance on arbitrary cutoffs like “95% variance explained” by offering a statistical technique to identify meaningful components, making dimensionality reduction more reliable and insightful in real-world data.
Speaker: Pradeep Bhadola, NAS-MUNA
Series 8:
| Sl. No | Date & Time | Title | Speaker & Afiliation | |
| 1 | 1st Sep 2025, ICT 14:00 hrs | TBA | Yossathorn Tawabutr, Chulalongkorn University |