Vatsal Sharan
Assistant Professor of Computer Science
Studies foundations of machine learning, including learning theory and memory-sample tradeoffs.
Vatsal Sharan's lab at USC works on learning theory & optimization. 31 papers since 2023, 25 at top venues, mostly NeurIPS and ICLR. 7 PhD students. 6 alumni. Applications close .
At a glance
- papers, 2023–2026
- 8/ year
- at T1 venues
- 81%
- PhD students
- 7
- MS students
- 0
- alumni listed
- 6
Papers by yearfrom the lab homepage, 2023–2026
- 11 papers2026 · T1 8 · collecting
- 7 papers2025 · T1 6
- 11 papers2024 · T1 9
- 2 papers2023 · T1 2
T1 venuesOtherVenues: NeurIPS, ICLR, ICML, COLT, AAAI
Research focuspaper titles that use each topic's words
- Learning theory & optimization3 papers
- studies foundations0 papers
- foundations0 papers
- learning theory0 papers
- theory0 papers
Paper list
From the lab homepage.
202611 papers
- STOC (T1)A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures
- AAAI (T1)An External Fairness Evaluation of LinkedIn Talent Search
- ITCSAuditability and the Landscape of Distance to Multicalibration
- COLMConvergent Evolution: How Different Language Models Learn Similar Number Representations
- COLT (T1)Efficient Swap Multicalibration of Elicitable Properties
- ICLR (T1)FoNE: Precise Single-Token Number Embeddings via Fourier Features
- ICLR (T1)How Muon's Spectral Design Benefits Generalization: A Study on Imbalanced Data
- ICLR (T1)Latent Concept Disentanglement in Transformer-based Language Models
- FORCLimitations on Safe, Trusted, Artificial General Intelligence
- ACL (T1)Textual Steering Vectors Can Improve Visual Understanding in Multimodal Large Language Models
- ICML (T1)Transformers Provably Learn Algorithmic Solutions for Graph Connectivity, But Only with the Right Data
20257 papers
- NeurIPS (T1)Discovering Data Structures: Nearest Neighbor Search and Beyond
- NeurIPS (T1)Improved Bounds for Swap Multicalibration and Swap Omniprediction
- AISTATS (T1)On the Inherent Privacy of Zeroth Order Projected Gradient Descent
- ALTProper Learnability and the Role of Unlabeled Data
- NeurIPS (T1)Simultaneous Swap Regret Minimization via KL-Calibration
- NeurIPS (T1)The Rich and the Simple: On the Implicit Bias of Adam and SGD
- ICLR (T1)Transformers Learn Low Sensitivity Functions: Investigations and Implications
202411 papers
- Journal of the ACM (T1)Efficient Convex Optimization Requires Superlinear Memory
- TMLR (T1)Mitigating Simplicity Bias in Deep Learning for Improved OOD Generalization and Robustness
- Annals of StatisticsOn the Statistical Complexity of Sample Amplification
- Open Problem @ COLTOpen Problem: Can Local Regularization Learn All Multiclass Problems?
- NeurIPS (T1)Optimal Multiclass U-Calibration Error and Beyond
- NeurIPS (T1)Pre-trained Large Language Models Use Fourier Features to Compute Addition
- COLT (T1)Regularization and Optimal Multiclass Learning
- ICML (T1)Stability and Multigroup Fairness in Ranking with Uncertain Predictions
- NeurIPS (T1)Transductive Sample Complexities Are Compact
- NeurIPS (T1)Transformers Learn to Achieve Second-Order Convergence Rates for In-Context Linear Regression
- NeurIPS (T1)When is Multicalibration Post-Processing Necessary?
20232 papers
- ICML (T1)Fairness in Matching under Uncertainty
- SIGMOD (T1)NeuroSketch: A Neural Network Method for Fast and Approximate Evaluation of Range Aggregate Queries
Where alumni went6 listed, 0% to industry
Industry 0Academia 5
Also to
- University of Washington
- Harvard University
- the University of Chicago
Apply to USCFall 2027
Computer Science PhD (Viterbi)
- Round
- Single round (Dec 15)
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- USStanford
Sanmi Koyejo
- Trustworthy & safe ML
- CAToronto
Roger Grosse
- Learning theory & optimization
- USGeorgia Tech
Jesse Thomason
- Multimodal & grounded language
- USCMU
Rayid Ghani
- Trustworthy & safe ML
- USUMD
Han Shao
- Trustworthy & safe ML
- GBImperial
Roberto Bondesan
- Learning theory & optimization
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
USCcloses