Relative Drawing Identification Complexity Is Invariant to Modality in Vision-Language Models
Abstract
Large language models have become multimodal, and many of them are said to integrate their modalities using common representations. If this were true, a drawing of a car as an image, for instance, should map to a similar area in the latent space as a textual description of the strokes that form the drawing. To explore this in a black-box access regime to these models, we propose the use of machine teaching, a theory that studies the minimal set of examples a teacher needs to choose so that the learner captures the concept. In this paper, we evaluate the complexity of teaching vision-language models a subset of objects in the Quick, Draw! dataset using two presentations: raw images as bitmaps and trace coordinates in TikZ format. The results indicate that image-based representations generally require fewer segments and achieve higher accuracy than coordinate-based representations. But, surprisingly, the teaching size usually ranks concepts similarly across both modalities, even when controlling for (a human proxy of) concept priors, suggesting that the simplicity of concepts may be an inherent property that transcends modality representations.
More information
- Authors
- Diogo Freitas; Brigt Håvardstun; Darío Garigliotti; Jan Arne Telle; Cèsar Ferri; José Hernández-Orallo
- Date
- 2025
- Conference
- 28th European Conference on Artificial Intelligence (ECAI 2025)
- Book
- Frontiers in Artificial Intelligence and Applications, Volume 413
- Pages
- 4507–4514
- Source
- Link
Citation
@inproceedings{freitas2025relative,
author = {Freitas, Diogo and Håvardstun, Brigt and Garigliotti, Darío and Telle, Jan Arne and Ferri, Cèsar and Hernández-Orallo, José},
booktitle = {ECAI 2025 - 28th European Conference on Artificial Intelligence},
doi = {10.3233/FAIA251351},
pages = {4507--4514},
series = {Frontiers in Artificial Intelligence and Applications},
title = {Relative Drawing Identification Complexity Is Invariant to Modality in Vision-Language Models},
volume = {413},
year = {2025}
}
