Semi-Supervised Optimal Transport for Heterogeneous Domain Adaptation

Yan, Yuguang; Wen, Li; Wu, Hanrui; Min, Huaqing; Tan, Mingkui; Wu, Qingyao

doi:10.24963/ijcai.2018/412

Cited by 84 publications

(54 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Metric Transportation on HEterogeneous RErepresentations (MAPHERE) [41] alternatively updates the asymmetric transformation and optimal transportation map. To preserve the cross-domain semantic consistency, Yan et al proposed Semi-supervised entropic Gromov-Wasserstein (SGW) [37], which incorporates label information with Entropic Gromov-Wasserstein to learn the transportation strategy.…”

Section: Related Workmentioning

confidence: 99%

Prototype-Matching Graph Network for Heterogeneous Domain Adaptation

Wang

Luo

Huang

et al. 2020

Proceedings of the 28th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Even though the multimedia data is ubiquitous on the web, the scarcity of the annotated data and variety of data modalities hinder their usage by multimedia applications. Heterogeneous domain adaptation (HDA) has therefore arisen to address such limitations by facilitating the knowledge transfer between heterogeneous domains. Existing HDA methods only focus on aligning the crossdomain feature distributions and ignore the importance of maximizing the margin among different classes, which may lead to a suboptimal classification performance. To tackle this problem, in this paper, we propose the Prototype-Matching Graph Network (PMGN), which gradually explores the domain-invariant class prototype representations. Specifically, we build an end-to-end Graph Prototypical Network, which computes the class prototypes through multiple layers of edge learning, node aggregation, and discrepancy minimization. Our framework utilizes the Swap training strategy to provide adequate supervision for training the edge learning component. Moreover, the proposed PMGN can be equipped with the clustering module that utilises the KL-divergence as a distance metric to reduce the distribution difference between the source and target data. Extensive experiments on three HDA tasks (i.e. object recognition, text-to-image classification, and text categorization) demonstrate the superiority of our approach over the state-of-theart HDA methods. CCS CONCEPTS • Computing methodologies → Transfer learning.

show abstract

Section: Related Workmentioning

confidence: 99%

Prototype-Matching Graph Network for Heterogeneous Domain Adaptation

Wang

Luo

Huang

et al. 2020

Proceedings of the 28th ACM International Conference on Multimedia

View full text Add to dashboard Cite

show abstract

“…For the heterogeneous approaches, which can construct a shared bridge on different feature spaces, the key idea is to map different feature spaces to a same latent one. Reference [37] developed a semi-supervised approach to match the examples and preserve the semantic consistency between heterogeneous domains. Reference [38] proposed a novel HTL approach (Deep-MCA) based on a structure with adversarial kernel training to obtain an end-toend solution.…”

Section: Related Workmentioning

confidence: 99%

Multi-Group Transfer Learning on Multiple Latent Spaces for Text Classification

Pan

Cui

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Transfer learning aims to leverage valuable information in one domain to promote the learning tasks in the other domain. Some recent studies indicated that the latent information, which has a close relationship with the high-level concepts, are more suitable for cross-domain text classification than learning raw features. To obtain more latent information existing in the latent feature space, some previous methods constructed multiple latent feature spaces. However, those methods ignored that the latent information of different latent spaces may lack the relevance for promoting the adaptability of transfer learning models, even may lead to negative knowledge transfer when there exists a glaring discrepancy among the different latent spaces. Additionally, since those methods learn the latent space distributions using a strategy of directpromotion, their computational complexity increases exponentially as the number of latent spaces increases. To tackle this challenge, this paper proposes a Multiple Groups Transfer Learning (MGTL) method. MGTL first constructs multiple different latent feature spaces and then integrates the adjacent ones that have a similar latent feature dimension into one latent space group. Along this way, multiple latent space groups can be obtained. To enhance the relevance among these latent space groups, MGTL makes the adjacent groups contain one same latent space at least. Then, different groups will have more relevance than raw latent spaces. Second, MGTL utilizes an indirect-promotion strategy to connect different latent space groups. The computational complexity of MGTL increases linearly as the number of latent space groups increases and is superior to those multiple latent space methods based on direct-promotion. In addition, an iterative algorithm is proposed to solve the optimization problem. Finally, a set of systematic experiments demonstrate that MGTL outperforms all the compared existing methods. INDEX TERMS Transfer learning, non-negative matrix tri-factorization, multi-group, cross-domain classification.

show abstract

“…Wu et al [36] proposed online transfer learning paradigms in which the source and target domains are leveraged adaptively. Yan et al [37] proposed a semi-supervised algorithm for heterogeneous domain adaptation by exploiting the theory of optimal transport. The method can also be used to find the optimal discriminative correlation subspace for the source and target data [38].…”

Section: B Heterogeneous Transfer Learning On Software Defect Predictionmentioning

confidence: 99%

Multiview Transfer Learning for Software Defect Prediction

Chen

Yang

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Most software defect prediction models usually assume that enough historical training instances with labels are available. Additionally, the training data and the predicted instances should share the same features to ensure the prediction accuracy. However, in practice, there are many datasets with different granularities containing information in different dimensions. Therefore, it is valuable to effectively use the small scale and different dimensions of data as training instances to improve the prediction performance of the model. We propose a heterogeneous data orienting multiview transfer learning for software defect prediction, denoted as MTDP, which can achieve different dimensions and granularities features to automatically learn labels through neural network models. With this multiview transfer method, lots of training instances are provided for software defect prediction model to ensure the effectiveness of training labels. The proposed MTDP method has four main stages: 1) build heterogeneous transfer models; 2) transfer heterogeneous instances to generate quasi-real instances; 3) label quasi-real instances through co-training and then expand the training set; and (4) construct improved software defect prediction models. The experimental results show that the quasi-real instances have similar effects compared with real instances. Moreover, the software defect prediction performance can be improved by introducing the quasi-real instances into the training dataset.INDEX TERMS Multiview transfer learning, software defect prediction, active learning, neural network.

show abstract

Semi-Supervised Optimal Transport for Heterogeneous Domain Adaptation

Cited by 84 publications

References 14 publications

Prototype-Matching Graph Network for Heterogeneous Domain Adaptation

Prototype-Matching Graph Network for Heterogeneous Domain Adaptation

Multi-Group Transfer Learning on Multiple Latent Spaces for Text Classification

Multiview Transfer Learning for Software Defect Prediction

Contact Info

Product

Resources

About