Adel Javanmard

Understanding the Role of Training Data in Test-Time Scaling

Self-Boost via Optimal Retraining: An Analysis via Approximate Message Passing

Retraining with Predicted Hard Labels Provably Increases Model Accuracy

Robust Feature Learning for Multi-Index Models in High Dimensions

Integer Programming for Generalized Causal Bootstrap Designs

Improving the Variance of Differentially Private Randomized Experiments through Clustering

DeepCrossAttention: Supercharging Transformer Residual Connections

Retraining with Predicted Hard Labels Provably Increases Model Accuracy