Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wang, Yinsong, Shahrampour, Shahin
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2302.02224
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929391358967808
author Wang, Yinsong
Shahrampour, Shahin
author_facet Wang, Yinsong
Shahrampour, Shahin
contents This paper addresses a cross-modal learning framework, where the objective is to enhance the performance of supervised learning in the primary modality using an unlabeled, unpaired secondary modality. Taking a probabilistic approach for missing information estimation, we show that the extra information contained in the secondary modality can be estimated via Nadaraya-Watson (NW) kernel regression, which can further be expressed as a kernelized cross-attention module (under linear transformation). This expression lays the foundation for introducing The Attention Patch (TAP), a simple neural network add-on that can be trained to allow data-level knowledge transfer from the unlabeled modality. We provide extensive numerical simulations using real-world datasets to show that TAP can provide statistically significant improvement in generalization across different domains and different neural network architectures, making use of seemingly unusable unlabeled cross-modal data.
format Preprint
id arxiv_https___arxiv_org_abs_2302_02224
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle TAP: The Attention Patch for Cross-Modal Knowledge Transfer from Unlabeled Modality
Wang, Yinsong
Shahrampour, Shahin
Machine Learning
This paper addresses a cross-modal learning framework, where the objective is to enhance the performance of supervised learning in the primary modality using an unlabeled, unpaired secondary modality. Taking a probabilistic approach for missing information estimation, we show that the extra information contained in the secondary modality can be estimated via Nadaraya-Watson (NW) kernel regression, which can further be expressed as a kernelized cross-attention module (under linear transformation). This expression lays the foundation for introducing The Attention Patch (TAP), a simple neural network add-on that can be trained to allow data-level knowledge transfer from the unlabeled modality. We provide extensive numerical simulations using real-world datasets to show that TAP can provide statistically significant improvement in generalization across different domains and different neural network architectures, making use of seemingly unusable unlabeled cross-modal data.
title TAP: The Attention Patch for Cross-Modal Knowledge Transfer from Unlabeled Modality
topic Machine Learning
url https://arxiv.org/abs/2302.02224