Lei Clifton

LLMs can lose over 30% of their accuracy in medical judgment when exposed to misleading contexts, revealing a critical vulnerability in their deployment for health advice.

Ditching noisy SGD trajectories for smooth Bezier curves unlocks better dataset condensation, especially when data is scarce.

Dataset condensation, previously limited to neural networks, can now democratize access to clinical data by enabling privacy-preserving training of classical models like decision trees and Cox regression.