A Generalization Theory for Zero-Shot Prediction

Ronak Mehta, Zaid Harchaoui
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:43603-43660, 2025.

Abstract

A modern paradigm for generalization in machine learning and AI consists of pre-training a task-agnostic foundation model, generally obtained using self-supervised and multimodal contrastive learning. The resulting representations can be used for prediction on a downstream task for which no labeled data is available. We present a theoretical framework to better understand this approach, called zero-shot prediction. We identify the target quantities that zero-shot prediction aims to learn, or learns in passing, and the key conditional independence relationships that enable its generalization ability.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-mehta25a, title = {A Generalization Theory for Zero-Shot Prediction}, author = {Mehta, Ronak and Harchaoui, Zaid}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {43603--43660}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/mehta25a/mehta25a.pdf}, url = {https://proceedings.mlr.press/v267/mehta25a.html}, abstract = {A modern paradigm for generalization in machine learning and AI consists of pre-training a task-agnostic foundation model, generally obtained using self-supervised and multimodal contrastive learning. The resulting representations can be used for prediction on a downstream task for which no labeled data is available. We present a theoretical framework to better understand this approach, called zero-shot prediction. We identify the target quantities that zero-shot prediction aims to learn, or learns in passing, and the key conditional independence relationships that enable its generalization ability.} }
Endnote
%0 Conference Paper %T A Generalization Theory for Zero-Shot Prediction %A Ronak Mehta %A Zaid Harchaoui %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-mehta25a %I PMLR %P 43603--43660 %U https://proceedings.mlr.press/v267/mehta25a.html %V 267 %X A modern paradigm for generalization in machine learning and AI consists of pre-training a task-agnostic foundation model, generally obtained using self-supervised and multimodal contrastive learning. The resulting representations can be used for prediction on a downstream task for which no labeled data is available. We present a theoretical framework to better understand this approach, called zero-shot prediction. We identify the target quantities that zero-shot prediction aims to learn, or learns in passing, and the key conditional independence relationships that enable its generalization ability.
APA
Mehta, R. & Harchaoui, Z.. (2025). A Generalization Theory for Zero-Shot Prediction. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:43603-43660 Available from https://proceedings.mlr.press/v267/mehta25a.html.

Related Material