2 papers
cs.CV2025
Refining Skewed Perceptions in Vision-Language Contrastive Models through Visual Representations
Haocheng Dai, Sarang Joshi
Large vision-language contrastive models (VLCMs), such as CLIP, have become foundational, demonstrating remarkable success across a variety of downstream tasks. Despite their advan…
cs.LG2025
Uncovering Memorization Effect in the Presence of Spurious Correlations
Chenyu You, Haocheng Dai, Yifei Min +3
Machine learning models often rely on simple spurious features -- patterns in training data that correlate with targets but are not causally related to them, like image backgrounds…