Transitive Transfer Learning

Tan, Ben; Song, Yangqiu; Zhong, Erheng; Yang, Qiang

doi:10.1145/2783258.2783295

Cited by 156 publications

(56 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Since we focus on binary classification tasks, we first select ten classes from the NUS-WIDE data set (each class contains 300 images) and then use images with two classes (e.g., river and sky) to conduct a binary classification task, where the data consist of images with either the positive class (e.g., river) or the negative class (e.g., sky). This setting is also used in the experiments reported in [11]- [13], [15]- [17], and [20]. The number of the generated tasks is 10 2 = 45.…”

Section: Methodsmentioning

confidence: 99%

Online Heterogeneous Transfer by Hedge Ensemble of Offline and Online Decisions

Yan

Tan

et al. 2017

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

In this paper, we study the online heterogeneous transfer (OHT) learning problem, where the target data of interest arrive in an online manner, while the source data and auxiliary co-occurrence data are from offline sources and can be easily annotated. OHT is very challenging, since the feature spaces of the source and target domains are different. To address this, we propose a novel technique called OHT by hedge ensemble by exploiting both offline knowledge and online knowledge of different domains. To this end, we build an offline decision function based on a heterogeneous similarity that is constructed using labeled source data and unlabeled auxiliary co-occurrence data. After that, an online decision function is learned from the target data. Last, we employ a hedge weighting strategy to combine the offline and online decision functions to exploit knowledge from the source and target domains of different feature spaces. We also provide a theoretical analysis regarding the mistake bounds of the proposed approach. Comprehensive experiments on three real-world data sets demonstrate the effectiveness of the proposed technique.

show abstract

Section: Methodsmentioning

confidence: 99%

Online Heterogeneous Transfer by Hedge Ensemble of Offline and Online Decisions

Yan

Tan

et al. 2017

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

show abstract

“…Different supervised, unsupervised and semi-supervised methods have been proposed for a wide variety of applications such as image classification [21], WiFi-localization on time variant data [13], and web document classification [5,12]. Recently transfer learning is successfully used to classify images by learning the classifier on related text data [16,20]. Transfer learning is broadly classified into inductive, transductive and unsupervised transfer learning [14].…”

Section: Related Workmentioning

confidence: 99%

User-guided Cross-domain Sentiment Classification

Nelakurthi

Tong

Maciejewski³

et al. 2017

Proceedings of the 2017 SIAM International Conference on Data Mining

View full text Add to dashboard Cite

Sentiment analysis has been studied for decades, and it is widely used in many real applications such as media monitoring. In sentiment analysis, when addressing the problem of limited labeled data from the target domain, transfer learning, or domain adaptation, has been successfully applied, which borrows information from a relevant source domain with abundant labeled data to improve the prediction performance in the target domain. The key to transfer learning is how to model the relatedness among different domains.For sentiment analysis, a common practice is to assume similar sentiment polarity for the common keywords shared by different domains. However, existing methods largely overlooked the human factor, i.e., the users who expressed such sentiment. In this paper, we address this problem by explicitly modeling the human factor related to sentiment classification. In particular, we assume that the content generated by the same user across different domains is biased in the same way in terms of the sentiment polarity. In other words, optimistic/pessimistic users demonstrate consistent sentiment patterns, no matter what the context is. To this end, we propose a new graph-based approach named U-Cross, which models the relatedness of different domains via both the shared users and keywords. It is non-parametric and semi-supervised in nature. Furthermore, we also study the problem of shared user selection to prevent 'negative transfer'. In the experiments, we demonstrate the effectiveness of U-Cross by comparing it with existing state-of-the-art techniques on three real data sets.

show abstract

“…Transfer learning [21], a kind of method solves the question that how to take advantage of limited data, it can transfer knowledge learned from the existing data to help the studying in the future. The purpose of transfer learning is to help learning tasks in a new environment by knowledge learned from an available environment, so it won't make the same distribution assumption, as machine learning does.…”

Section: Introductionsmentioning

confidence: 99%