Large-scale semantic web image retrieval using bimodal deep learning techniques

Huang, Changqin; Xu, Haijiao; Xie, Liang; Zhu, Jia; Xu, Chunyan; Tang, Yong

doi:10.1016/j.ins.2017.11.043

Cited by 15 publications

(6 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Thirdly, we could analyse the quality of the Web images using a multi-dimensional image quality prediction model [42] , and propose as candidates only the images deemed of high quality. In addition, we could extract image semantic features [11] and match these against Web page text or genre [23] to further narrow down the candidates. For representing the text of a Web page one can make use of natural language processing techniques for normalization and summarization.…”

Section: Resultsmentioning

confidence: 99%

Determining the most representative image on a Web page

Vyas

Frăsincar

2020

Information Sciences

View full text Add to dashboard Cite

We investigate how to determine the most representative image on a Web page. This problem has not been thoroughly investigated and, up to today, only expert-based algorithms have been proposed in the literature. We attempt to improve the performance of known algorithms with the use of Support Vector Machines (SVM). Besides, our algorithm distinguishes itself from existing literature with the introduction of novel image features, including previously unused meta-data protocols. Also, we design and attempt a less-restrictive ranking methodology in the image preprocessing stage of our algorithm. We find that the application of the SVM framework with our improved classification methodology increases the F 1 score from 27.2% to 38.5%, as compared to a state-of-the-art method. Introducing novel image features and applying backward feature selection, we find that the F 1 score rises to 40.0%. Lastly, we use a class-weighted SVM in order to resolve the imbalance in number of representative images. This final modification improves the classification performance of our algorithm even further to 43.9%, outperforming our benchmark algorithms, including those of Facebook and Google. Suggested beneficiaries are the search engine community, image retrieval community, including the commercial sector due to superior performance.

show abstract

Section: Resultsmentioning

confidence: 99%

Determining the most representative image on a Web page

Vyas

Frăsincar

2020

Information Sciences

View full text Add to dashboard Cite

show abstract

“…This task facilities the computer‐human interaction as it simplifies the task of retrieval by restricting the search for similar images to a smaller domain of the dataset. Recently, researchers developed probabilistic models for organizing image collections into structures that can be used for indexing, browsing, retrieval and summarization . For instance, the authors in Reference proposed an indexing schema specifically designed for image collections, based on image content and semantics.…”

Section: Resultsmentioning

confidence: 99%

Proportional data modeling via selection and estimation of a finite mixture of scaled Dirichlet distributions

Zamzami

Alsuroji

Eromonsele

et al. 2019

Computational Intelligence

View full text Add to dashboard Cite

This paper proposes an unsupervised algorithm for learning a finite mixture of scaled Dirichlet distributions. Parameters estimation is based on the maximum likelihood approach, and the minimum message length (MML) criterion is proposed for selecting the optimal number of components. This research work is motivated by the flexibility issues of the Dirichlet distribution, the widely used model for multivariate proportional data, which has prompted a number of scholars to search for generalizations of the Dirichlet. By introducing the extra parameters of the scaled Dirichlet, several useful statistical models could be obtained. Experimental results are presented using both synthetic and real datasets. Moreover, challenging real‐world applications are empirically investigated to evaluate the efficiency of our proposed statistical framework.

show abstract

“…Los resultados demostraron que las pruebas de la red visual tienen un alto índice de clasificación y asignación al nombrar correctamente las imágenes. Huang et al [14] planteó que el MRBDL (Multi-concept Retrieval using Bimodal Deep Learning) capturó correlaciones semánticas entre una imagen visual con sus etiquetas libres contextuales.…”

Section: Técnicas Deep Learning En El Desarrollo De Softwareunclassified

Proceso de desarrollo de aplicaciones para el dominio social media usando técnicas de Deep Learning

Morales¹,

Morales²,

Alor-Hernández³

et al. 2019

RCS

View full text Add to dashboard Cite

Resumen. Los avances del Deep Learning, son cada vez más notorios en el desarrollo de software. Sin embargo, a pesar de tener un gran avance aún no se explotan las técnicas de Deep Learning para los generadores automáticos de software, con atención especial en el desarrollo de aplicaciones Social Media. Por lo tanto, se propone un proceso de generación automática de aplicaciones basado en técnicas de Deep Learning, a partir de una imagen digital que representa una interfaz de usuario. Para lo cual se deben identificar UIDPs (User Interface Design Patterns) válidos para interfaces de aplicaciones de Social Media a través de técnicas de aprendizaje profundo. Las ventajas de este trabajo es el enriquecer las aplicaciones, reducir costos y mejorar el desempeño. Por último, se presenta un caso de estudio que muestra la implementación de las técnicas de Deep Learning en la identificación de UIDPs.Palabras clave: generación automática de software, procesamiento de imágenes, proceso de desarrollo, aprendizaje profundo, social media, UIDPs.

show abstract

Large-scale semantic web image retrieval using bimodal deep learning techniques

Cited by 15 publications

References 16 publications

Determining the most representative image on a Web page

Determining the most representative image on a Web page

Proportional data modeling via selection and estimation of a finite mixture of scaled Dirichlet distributions

Proceso de desarrollo de aplicaciones para el dominio social media usando técnicas de Deep Learning

Contact Info

Product

Resources

About