Abstract
Multi-domain learning aims to benefit from simultaneously learning across several different but related domains. In this chapter, we propose a single framework that unifies multi-domain learning (MDL) and the related but better studied area of multitask learning (MTL). By exploiting the concept of a semantic descriptor we show how our framework encompasses various classic and recent MDL/MTL algorithms as special cases with different semantic descriptor encodings. As a second contribution, we present a higher order generalization of this framework, capable of simultaneous multitask-multi-domain learning. This generalization has two mathematically equivalent views in multilinear algebra and gated neural networks, respectively. Moreover, by exploiting the semantic descriptor, it provides neural networks the capability of zero-shot learning (ZSL), where a classifier is generated for an unseen class without any training data; as well as zero-shot domain adaptation (ZSDA), where a model is generated for an unseen domain without any training data. In practice, this framework provides a powerful yet easy to implement method that can be flexibly applied to MTL, MDL, ZSL, and ZSDA.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
All vectors (feature \(\mathbf {x}\), weights \(\mathbf {w}\), and domain descriptor \(\mathbf {z}\)) are by default column vectors.
- 2.
Note that, in any multi-domain or multitask learning problem, all instances are at least implicitly associated with a semantic descriptor indicating their domain (task).
- 3.
RMTL: Regularized Multi–Task Learning, FEDA: Frustratingly Easy Domain Adaptation, MTFL: Multi–Task Feature Learning, TNMTL: Trace-Norm Multi–Task Learning, and GO-MTL: Grouping and Overlap for Multi–Task Learning.
- 4.
Note that despite the title, [41] actually considers unsupervised domain adaptation without target domain labels, but with target domain data.
- 5.
We omit the upper and lower scripts for clarity.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this chapter
Cite this chapter
Yang, Y., Hospedales, T.M. (2017). Unifying Multi-domain Multitask Learning: Tensor and Neural Network Perspectives. In: Csurka, G. (eds) Domain Adaptation in Computer Vision Applications. Advances in Computer Vision and Pattern Recognition. Springer, Cham. https://doi.org/10.1007/978-3-319-58347-1_16
Download citation
DOI: https://doi.org/10.1007/978-3-319-58347-1_16
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-58346-4
Online ISBN: 978-3-319-58347-1
eBook Packages: Computer ScienceComputer Science (R0)