Saved in:
Bibliographic Details
Main Authors: Mehta, Ronak, Harchaoui, Zaid
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.09128
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914014931451904
author Mehta, Ronak
Harchaoui, Zaid
author_facet Mehta, Ronak
Harchaoui, Zaid
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2507_09128
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Generalization Theory for Zero-Shot Prediction
Mehta, Ronak
Harchaoui, Zaid
Machine Learning
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.
title A Generalization Theory for Zero-Shot Prediction
topic Machine Learning
url https://arxiv.org/abs/2507.09128