Marzyeh Ghassemi

WRING Out The Bias: A Rotation-Based Alternative To Projection Debiasing

When Style Breaks Safety: Defending LLMs Against Superficial Style Alignment

Complementing Self-Consistency with Cross-Model Disagreement for Uncertainty Quantification

TOGA: Trigger Optimization for Clean Data Ordering Backdoor Attack

Tiered Agentic Oversight: A Hierarchical Multi-Agent System for Healthcare Safety

Train Long, Think Short: Curriculum Learning for Efficient Reasoning

Learning the Wrong Lessons: Syntactic-Domain Spurious Correlations in Language Models

Aggregation Hides Out-of-Distribution Generalization Failures from Spurious Correlations

On Group Sufficiency Under Label Bias

KScope: A Framework for Characterizing the Knowledge Status of Language Models

An Investigation of Memorization Risk in Healthcare Foundation Models

LEMoN: Label Error Detection using Multimodal Neighbors

LEMoN: Label Error Detection using Multimodal Neighbors

Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple Interactions

SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe