Suzanne Petryk

8 papers · 2021–2025 · 6 conferences · across top CS/AI conferences

Achievements

🌍 Conference Polyglot (6) 🌈 Renaissance Researcher (5) 🌉 Interdisciplinary Bridge 🐣 Hot Topic Early Bird 🐝 Cross-Pollinator (15) 🔥 Unstoppable (5)

Conferences

ICLR (3) CVPR (1) ECCV (1) EMNLP (1) NAACL (1) WACV (1)

Top co-authors

Trevor Darrell (6) Joseph E. Gonzalez (5) Anna Rohrbach (3) Marcus Rohrbach (3) Spencer Whitehead (2) John Canny (2) Joseph Gonzalez (2) Anish Kachinthaya (2) David M. Chan (2) Lisa Dunlap (2)

Keywords

multimodal learning (2) image captioning (2) large language model (2) zero-shot learning (2) semantic similarity (1) vision-language model (1) hallucination detection (1) object hallucination (1) compositional reasoning (1) confidence estimation (1) worst-group accuracy (1) token-level prediction (1) image caption evaluation (1) semantic relevance (1) hungarian matching (1) language specification (1) attention supervision (1) fine-grained classification (1) visual attention (1) human judgment (1)

Papers

Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations ICLR 2025

ALOHa: A New Measure for Hallucination in Captioning Models NAACL 2024

Simple Token-Level Confidence Improves Caption Correctness WACV 2024

CLAIR: Evaluating Image Captions with Large Language Models EMNLP 2023

On Guiding Visual Attention With Language Specification CVPR 2022

Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly ECCV 2022

Remembering for the Right Reasons: Explanations Reduce Catastrophic Forgetting ICLR 2021

NBDT: Neural-Backed Decision Tree ICLR 2021