Graph-based semi-supervised learning with multi-modality propagation for large-scale image datasets

Lee, Wen-Yu; Hsieh, Liang-Chi; Wu, Guan-Long; Hsu, Winston H.

doi:10.1016/j.jvcir.2012.12.002

Cited by 17 publications

(5 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They consist of two steps: graph construction and label inference and can be further structured in each step. Details can be found in the survey paper 36 .…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Semi-supervised classifier guided by discriminator

Jamroziński

Markowska–Kaczmar

2022

Sci Rep

View full text Add to dashboard Cite

Some machine learning applications do not allow for data augmentation or are applied to modalities where the augmentation is difficult to define. Our study aimed to develop a new method in semi-supervised learning (SSL) applicable to various modalities of data (images, sound, text), especially when augmentation is hard or impossible to define, i.e., medical images. Assuming that all samples, labeled and unlabeled, come from the same data distribution, we can say that labeled and unlabeled data sets used in the semi-supervised learning tasks are similar. Based on this observation, the data embeddings created by the classifier should also be similar for both sets. In our method, finding these embeddings is achieved based on two models—classifier and an auxiliary discriminator model, inspired by the Generative Adversarial Network (GAN) learning process. The classifier is trained to build embeddings for labeled and unlabeled datasets to cheat discriminator, which recognizes whether the embedding comes from a labeled or unlabeled dataset. The method was named the DGSSC from Discriminator Guided Semi-Supervised Classifier. The experimental research aimed evaluation of the proposed method on the classification task in combination with the teacher-student approach and comparison with other SSL methods. In most experiments, training the networks with the DGSSC method improves accuracy with the teacher-student approach. It does not deteriorate the accuracy of any experiment.

show abstract

“…They consist of two steps: graph construction and label inference and can be further structured in each step. Details can be found in the survey paper 36 .…”

Section: Related Workmentioning

confidence: 99%

“…Our literature survey shows that SSL methods are mainly developed using one data modality, i.e., images ( [13][14][15][36][37][38] ), text ( [39][40][41][42][43] ) or sound ( 12,44 ). This unwritten standard runs the risk that developed approaches are effective only for this type of data.…”

Section: Related Workmentioning

confidence: 99%

Semi-supervised classifier guided by discriminator

Jamroziński

Markowska–Kaczmar

2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Zhu et al [24] first leveraged label propagation for a graph based semi-supervised learning algorithm, and Zhu et al [25] further proposed an iterative algorithm from a close form solution for label propagation. After that label propagation has been applied to many domains, such as multimedia including image and video data [26] and information retrieval like relevance and keyword search [27], [28]. In addition, Rao and Yarowsky [29] proposed a parallel label propagation algorithm under the MapReducde framework.…”

Section: Related Workmentioning

confidence: 99%

Large-Scale Gender/Age Prediction of Tumblr Users

Zhang

et al. 2019

2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA)

View full text Add to dashboard Cite

Tumblr, as a leading content provider and social media, attracts 371 million monthly visits, 280 million blogs and 53.3 million daily posts 1 .The popularity of Tumblr provides great opportunities for advertisers to promote their products through sponsored posts. However, it is a challenging task to target specific demographic groups for ads, since Tumblr does not require user information like gender and ages during their registration. Hence, to promote ad targeting, it is essential to predict user's demography using rich content such as posts, images and social connections. In this paper, we propose graph based and deep learning models for age and gender predictions, which take into account user activities and content features. For graph based models, we come up with two approaches, network embedding and label propagation, to generate connection features as well as directly infer user's demography. For deep learning models, we leverage convolutional neural network (CNN) and multilayer perceptron (MLP) to prediction users' age and gender. Experimental results on real Tumblr daily dataset, with hundreds of millions of active users and billions of following relations, demonstrate that our approaches significantly outperform the baseline model, by improving the accuracy relatively by 81% for age, and the AUC and accuracy by 5% for gender.

show abstract

“…Wang et al [135] proposed a multi-graph based semisupervised learning technique that can incorporate multiple modalities, and multiple distance functions in the task of video annotation. Lee et al [75] proposed a new graph-based multi-label propagation technique and applied it to large datasets by utilizing a map-reduce framework. Though these methods can handle multi-view data, they fail to address the scenarios where classes/labels are arranged in a hierarchy and inference needs to be done following certain constraints between these classes.…”

Section: Hierarchical Semi-supervised Learningmentioning

confidence: 99%

Constrained Semi-supervised Learning in the Presence of Unanticipated Classes

Dalvi

2016

SIGIR Forum

View full text Add to dashboard Cite

Traditional semi-supervised learning (SSL) techniques consider the missing labels of unlabeled datapoints as latent/unobserved variables, and model these variables, and the parameters of the model, using techniques like Expectation Maximization (EM). Such semisupervised learning techniques are widely used for Automatic Knowledge Base Construction (AKBC) tasks. We consider two extensions to traditional SSL methods which make it more suitable for a variety of AKBC tasks. First, we consider jointly assigning multiple labels to each instance, with a flexible scheme for encoding constraints between assigned labels: this makes it possible, for instance, to assign labels at multiple levels from a hierarchy. Second, we account for another type of latent variable, in the form of unobserved classes. In open-domain webscale information extraction problems, it is an unrealistic assumption that the class ontology or topic hierarchy we are using is complete. Our proposed framework combines structural search for the best class hierarchy with SSL, reducing the semantic drift associated with erroneously grouping unanticipated classes with expected classes. Together, these extensions allow a single framework to handle a large number of knowledge extraction tasks, including macro-reading, noun-phrase classification, word sense disambiguation, alignment of KBs to wikipedia or on-line glossaries, and ontology extension. To summarize, this thesis argues that many AKBC tasks which have previously been addressed separately can be viewed as instances of single abstract problem: multiview semisupervised learning with an incomplete class hierarchy. In this thesis we present a generic EM framework for solving this abstract task.

show abstract

Graph-based semi-supervised learning with multi-modality propagation for large-scale image datasets

Cited by 17 publications

References 20 publications

Semi-supervised classifier guided by discriminator

Semi-supervised classifier guided by discriminator

Large-Scale Gender/Age Prediction of Tumblr Users

Constrained Semi-supervised Learning in the Presence of Unanticipated Classes

Contact Info

Product

Resources

About