Actions

Download Analytics Citations

Export to: EndNote  |  Zotero  |  Mendeley

Collections

This file is not currently in any collections.

Cross-Domain Structure Preserving Projection for Heterogeneous Domain Adaptation [dataset] Open Access

Heterogeneous Domain Adaptation (HDA) addresses the transfer learning problems where data from the source and target domains are of different modalities (e.g., texts and images) or feature dimensions (e.g., features extracted with different methods). It is useful for multi-modal data analysis. Traditional domain adaptation algorithms assume that the representations of source and target samples reside in the same feature space, hence are likely to fail in solving the heterogeneous domain adaptation problem. Contemporary state-of-the-art HDA approaches are usually composed of complex optimization objectives for favourable performance and are therefore computationally expensive and less generalizable. To address these issues, we propose a novel Cross-Domain Structure Preserving Projection (CDSPP) algorithm for HDA. As an extension of the classic LPP to heterogeneous domains, CDSPP aims to learn domain-specific projection to map sample features from source and target domains into a common subspace such that the class consistency is preserved and data distributions are sufficiently aligned. CDSPP is simple and has deterministic solutions by solving a generalized eigenvalue problem. It is naturally suitable for supervised HDA but has also been extended for semi-supervised HDA where the unlabelled target domain samples are available. Extensive experiments have been conducted on commonly used benchmark datasets (i.e. Office-Caltech, Multilingual Reuters Collection, NUS-WIDE-ImageNet) for HDA as well as the Office-Home dataset firstly introduced for HDA by ourselves due to its significantly larger number of classes than the existing ones (65 vs 10, 6 and 8). The experimental results of both supervised and semi-supervised HDA demonstrate the superior performance of our proposed method against contemporary state-of-the-art methods. Supporting dataset for: Cross-Domain Structure Preserving Projection for Heterogeneous Domain Adaptation (Q. Wang, T.P. Breckon), In Pattern Recognition, Elsevier, Volume 123, 2022. DOI: https://doi.org/10.1016/j.patcog.2021.108362 DRO: https://dro.dur.ac.uk/34127/

Descriptions

Resource type

Dataset

Contributors

Data curator: Wang, Qian ¹
Creator: Breckon, Toby ¹
Editor: Breckon, Toby ¹

¹ Durham University, UK

Funder

Durham University, UK

Research methods

empirical machine learning

Other description

Supporting open source software:  https://github.com/hellowangqian/cdspp-hda

Keyword

heterogeneous domain adaptation
cross-domain projection
image classification
text classification
machine learning

Subject

Neural networks (Computer science)
Artificial intelligence

Location

Durham, England, United Kingdom

Language

English

Cited in

doi:10.1016/j.patcog.2021.108362

Identifier

ark:/32150/r2jw827b67n
doi:10.15128/r2jw827b67n

Rights

Creative Commons Attribution 4.0 International (CC BY)

Publisher

Durham University

Date Created

February 2022

File Details

Depositor

T. Breckon

Date Uploaded

30 May 2022, 22:05:04

Date Modified

31 May 2022, 14:05:10

Audit Status

Audits have not yet been run on this file.

Characterization

File format: zip (ZIP Format)

Mime type: application/zip

File size: 1458559618

Last modified: 2022:05:30 23:50:48+01:00

Filename: UDA-features.zip

Original checksum: e90e26121a4df03bca56af61c9300b95