Pratik Patil

"Throughout my academic career, I had given some pretty good talks. But being considered the best speaker in the computer science department is like being known as the tallest of the Seven Dwarfs." — Randy Pausch, The Last Lecture

Talks

• Asymptotically Free Sketching and Applications in Ridge Regression

• Revisiting Model Optimism and Model Complexity in the Wake of Overparameterized Learning

• Mitigating Multiple Descents: A Model-Agnostic Framework for Risk Monotonization

• Subsampling, Ensembling, and Ridge Regression

• Overparameterization in Linear Models, and Uniform Consistency of Cross-Validation for Ridge Regression

• Objective Frequentist Uncertainty Quantification for Atmospheric CO2 Retrievals

• Bagging Regularized M-Estimators: Precise Asymptotics and Cross-Validation

• Cross-Validation for Finite Randomized Ensembles

• Frequentist Confidence Intervals for Functional in Constrained Inverse Problems

• Uplink-Downlink Duality between Multiple-Access and Broadcast Channels with Compressing Relays

• Layered Constructions for Low-Delay Streaming Codes