Rajiv Khanna
Engineering · Purdue University West Lafayette
Publications
56
Citations
1,192
Est. group size
—
Recurring co-author estimate
Active years
36
Publishing since 1990
Rajiv Khanna works on the mathematical foundations of machine learning, focusing on how training algorithms and data selection methods behave and why they generalize (or fail to). His work covers topics like data summarization (coresets), understanding optimizer stability (e.g., SGD, SAM), robustness of deep networks, and efficient methods for large-scale computation such as distributed learning and matrix approximation. This research is largely theoretical/methodological, aiming to explain and improve the reliability and efficiency of machine learning systems.
Publication output was relatively high around 2017-2020, dropped sharply in 2021-2023, and has shown a modest uptick in 2024-2025.
Generated by claude-sonnet-5 from public bibliographic data · Jul 20, 2026
- On the Support Vector Effect in DNNs: Rethinking Data Selection and Attribution
2025
- Structure-Aware Spectral Sparsification via Uniform Edge Sampling
arXiv (Cornell University) · 2025
- Stable Coresets via Posterior Sampling: Aligning Induced and Full Loss Landscapes
arXiv (Cornell University) · 2025
- A Unified Stability Analysis of SAM vs SGD: Role of Data Coherence and Emergence of Simplicity Bias
arXiv (Cornell University) · 2025
- A Precise Characterization of SGD Stability Using Loss Surface Geometry
arXiv (Cornell University) · 2024
- Approximating Memorization Using Loss Surface Geometry for Dataset Pruning and Summarization
2024
- Membership Privacy Risks of Sharpness Aware Minimization
arXiv (Cornell University) · 2023
- LocalNewton: Reducing Communication Bottleneck for Distributed Learning
arXiv (Cornell University) · 2021
- Improved Guarantees and a Multiple-descent Curve for Column Subset Selection and the Nystrom Method (Extended Abstract)
2021
- Generalization Properties of Stochastic Optimizers via Trajectory Analysis.
arXiv (Cornell University) · 2021
- Adversarially-Trained Deep Nets Transfer Better
arXiv (Cornell University) · 2020
- Boundary thickness and robustness in learning models
arXiv (Cornell University) · 2020
- Improved guarantees and a multiple-descent curve for Column Subset\n Selection and the Nystr\\"om method
arXiv (Cornell University) · 2020
- Adversarially-Trained Deep Nets Transfer Better: Illustration on Image Classification
arXiv (Cornell University) · 2020
- Improved guarantees and a multiple-descent curve for Column Subset Selection and the Nyström method
arXiv (Cornell University) · 2020
- arXiv (Cornell University)×30
- International Conference on Artificial Intelligence and Statistics×4
- Society for Industrial and Applied Mathematics eBooks×2
- Neural Information Processing Systems×1
- The Annals of Statistics×1
- Chih-Hao Fang
Engineering · Purdue University West Lafayette
- Petros Drineas
Engineering · Purdue University West Lafayette
- Kananart Kuwaranancharoen
Engineering · Purdue University West Lafayette
- Yuchen Zhang
Engineering · Purdue University West Lafayette
- Zhihui Zhu
Engineering · The Ohio State University
This profile was generated automatically from public scholarly data (OpenAlex). Group size and activity levels are estimates derived from co-authorship patterns.
Last updated Jul 20, 2026.
Claim or correct this profile