Joint Semantic and Latent Attribute Modelling for Cross-Class Transfer Learning
- 6 July 2017
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 40 (7) , 1625-1638
Abstract
A number of vision problems such as zero-shot learning and person re-identification can be considered as cross-class transfer learning problems. As mid-level semantic properties shared cross different object classes, attributes have been studied extensively for knowledge transfer across classes. Most previous attribute learning methods focus only on human-defined/nameable semantic attributes, whilst ignoring the fact there also exist undefined/latent shareable visual properties, or latent attributes. These latent attributes can be either discriminative or non-discriminative parts depending on whether they can contribute to an object recognition task. In this work, we argue that learning the latent attributes jointly with user-defined semantic attributes not only leads to better representation but also helps semantic attribute prediction. A novel dictionary learning model is proposed which decomposes the dictionary space into three parts corresponding to semantic, latent discriminative and latent background attributes respectively. Such a joint attribute learning model is then extended by following a multi-task transfer learning framework to address a more challenging unsupervised domain adaptation problem, where annotations are only available on an auxiliary dataset and the target dataset is completely unlabelled. Extensive experiments show that the proposed models, though being linear and thus extremely efficient to compute, produce state-of-the-art results on both zero-shot learning and person re-identification.Keywords
Funding Information
- National Natural Science Foundation of China (61390515, U1611461, 61425025, 61471042)
- National Basic Research Program of China (2015CB351806)
This publication has 73 references indexed in Scilit:
- Learning Multimodal Latent AttributesIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- Towards Person Identification and Re-identification with Attributes2012
- Disentangling Factors of Variation for Facial Expression Recognition2012
- Attribute Discovery via Predictable Discriminative Binary Codes2012
- Relaxed Pairwise Learned Metric for Person Re-identification2012
- Augmented Attribute Representations2012
- Unsupervised Discovery of Mid-Level Discriminative Patches2012
- Person Re-identification by Descriptive and Discriminative Classification2011
- $rm K$-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse RepresentationIEEE Transactions on Signal Processing, 2006
- Dictionary Learning Algorithms for Sparse RepresentationNeural Computation, 2003