Is CLIP ideal? No. Can we fix it? Yes!

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kang, Raphi, Song, Yue, Gkioxari, Georgia, Perona, Pietro
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911516495708160
author Kang, Raphi
Song, Yue
Gkioxari, Georgia
Perona, Pietro
author_facet Kang, Raphi
Song, Yue
Gkioxari, Georgia
Perona, Pietro
contents Contrastive Language-Image Pre-Training (CLIP) is a popular method for learning multimodal latent spaces with well-organized semantics. Despite its wide range of applications, CLIP's latent space is known to fail at handling complex visual-textual interactions. Recent works attempt to address its shortcomings with data-centric or algorithmic approaches. But what if the problem is more fundamental, and lies in the geometry of CLIP? Toward this end, we rigorously analyze CLIP's latent space properties, and prove that no CLIP-like joint embedding space exists which can correctly do any two of the following at the same time: 1. represent basic descriptions and image content, 2. represent attribute binding, 3. represent spatial location and relationships, 4. represent negation. Informed by this analysis, we propose Dense Cosine Similarity Maps (DCSMs) as a principled and interpretable scoring method for CLIP-like models, which solves the fundamental limitations of CLIP by retaining the semantic topology of the image patches and text tokens. This method improves upon the performance of classical CLIP-like joint encoder models on a wide array of benchmarks. We share our code and data here for reproducibility: https://github.com/Raphoo/DCSM_Ideal_CLIP
format Preprint
id arxiv_https___arxiv_org_abs_2503_08723
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Is CLIP ideal? No. Can we fix it? Yes!
Kang, Raphi
Song, Yue
Gkioxari, Georgia
Perona, Pietro
Machine Learning
Computer Vision and Pattern Recognition
Contrastive Language-Image Pre-Training (CLIP) is a popular method for learning multimodal latent spaces with well-organized semantics. Despite its wide range of applications, CLIP's latent space is known to fail at handling complex visual-textual interactions. Recent works attempt to address its shortcomings with data-centric or algorithmic approaches. But what if the problem is more fundamental, and lies in the geometry of CLIP? Toward this end, we rigorously analyze CLIP's latent space properties, and prove that no CLIP-like joint embedding space exists which can correctly do any two of the following at the same time: 1. represent basic descriptions and image content, 2. represent attribute binding, 3. represent spatial location and relationships, 4. represent negation. Informed by this analysis, we propose Dense Cosine Similarity Maps (DCSMs) as a principled and interpretable scoring method for CLIP-like models, which solves the fundamental limitations of CLIP by retaining the semantic topology of the image patches and text tokens. This method improves upon the performance of classical CLIP-like joint encoder models on a wide array of benchmarks. We share our code and data here for reproducibility: https://github.com/Raphoo/DCSM_Ideal_CLIP
title Is CLIP ideal? No. Can we fix it? Yes!
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.08723