Data-driven two-layer visual dictionary structure learning

Xiangchun Yu, Zhezhou Yu (Corresponding Author), Lei Wu, Wei Pang, Chenghua Lin

Research output: Contribution to journalArticle

Abstract

An important issue in statistical modeling is to determine the complexity of the model based on the scale of data so as to effectively mitigate the model’s overfitting problems without big data. We adopt a data-driven approach to automatically determine the number of components of the model. In order to better extract robust features, we propose a framework of data-driven two-layer structure visual dictionary learning (DTSVDL). It works by dividing the visual dictionary structure learning into two levels: the attribute layer and the detail layer. In the attribute layer, the attributes of the image dataset are learned, and these attributes are obtained by a data-driven Bayesian nonparametric model. Then, in the detail layer, the detailed information over attributes is further explored and refined, and the attributes are weighted by the number of effective observations associated with each attribute. Our proposed approach has three main advantages: (1) the two-layer structure makes our building visual dictionary be more expressive; (2) the number of components in the attribute layer can be determined automatically from the data; (3) the components are automatically determined based on the scale of visual words; therefore, our model can well mitigate the overfitting problem. In addition, by comparing with stacked autoencoders, stacked denoising autoencoders, LeNet-5, speeded-up robust features, and pretrained deep learning model ImageNet-VGG-F algorithms, we find that our approach achieves satisfactory image categorization results on two benchmark datasets. Specifically, higher categorization performance is achieved than by the classical approaches on 15 scene categories and action datasets. We conclude that the resulting DTSVDL possesses a good generality derived from attribute information as well as an excellent distinction derived from detailed information. In other words, the visual dictionary learned by our algorithm is more expressive and discriminatory.
Original languageEnglish
Article number023006
JournalJournal of Electronic Imaging
Volume28
Issue number2
DOIs
Publication statusPublished - 8 Mar 2019

Fingerprint

Glossaries

Keywords

  • statistical modeling
  • over-fitting
  • visual dictionary
  • Bayesian nonparametric model
  • deep learning
  • overfitting
  • HIERARCHICAL MODEL
  • WORDS
  • BAG
  • LATENT DIRICHLET ALLOCATION
  • FEATURES

Cite this

Data-driven two-layer visual dictionary structure learning. / Yu, Xiangchun; Yu, Zhezhou (Corresponding Author); Wu, Lei; Pang, Wei; Lin, Chenghua.

In: Journal of Electronic Imaging, Vol. 28, No. 2, 023006, 08.03.2019.

Research output: Contribution to journalArticle

Yu, Xiangchun ; Yu, Zhezhou ; Wu, Lei ; Pang, Wei ; Lin, Chenghua. / Data-driven two-layer visual dictionary structure learning. In: Journal of Electronic Imaging. 2019 ; Vol. 28, No. 2.
@article{8282aa38c216488ebe307e75d2e79b61,
title = "Data-driven two-layer visual dictionary structure learning",
abstract = "An important issue in statistical modeling is to determine the complexity of the model based on the scale of data so as to effectively mitigate the model’s overfitting problems without big data. We adopt a data-driven approach to automatically determine the number of components of the model. In order to better extract robust features, we propose a framework of data-driven two-layer structure visual dictionary learning (DTSVDL). It works by dividing the visual dictionary structure learning into two levels: the attribute layer and the detail layer. In the attribute layer, the attributes of the image dataset are learned, and these attributes are obtained by a data-driven Bayesian nonparametric model. Then, in the detail layer, the detailed information over attributes is further explored and refined, and the attributes are weighted by the number of effective observations associated with each attribute. Our proposed approach has three main advantages: (1) the two-layer structure makes our building visual dictionary be more expressive; (2) the number of components in the attribute layer can be determined automatically from the data; (3) the components are automatically determined based on the scale of visual words; therefore, our model can well mitigate the overfitting problem. In addition, by comparing with stacked autoencoders, stacked denoising autoencoders, LeNet-5, speeded-up robust features, and pretrained deep learning model ImageNet-VGG-F algorithms, we find that our approach achieves satisfactory image categorization results on two benchmark datasets. Specifically, higher categorization performance is achieved than by the classical approaches on 15 scene categories and action datasets. We conclude that the resulting DTSVDL possesses a good generality derived from attribute information as well as an excellent distinction derived from detailed information. In other words, the visual dictionary learned by our algorithm is more expressive and discriminatory.",
keywords = "statistical modeling, over-fitting, visual dictionary, Bayesian nonparametric model, deep learning, overfitting, HIERARCHICAL MODEL, WORDS, BAG, LATENT DIRICHLET ALLOCATION, FEATURES",
author = "Xiangchun Yu and Zhezhou Yu and Lei Wu and Wei Pang and Chenghua Lin",
note = "This work was supported by (1) the Science and Technology Developing Project of Jilin Province, China (Grant No. 20150204007GX) and (2) the Key Laboratory of Symbolic computation and Knowledge Engineering, Ministry of Education.",
year = "2019",
month = "3",
day = "8",
doi = "10.1117/1.JEI.28.2.023006",
language = "English",
volume = "28",
journal = "Journal of Electronic Imaging",
issn = "1017-9909",
publisher = "SPIE",
number = "2",

}

TY - JOUR

T1 - Data-driven two-layer visual dictionary structure learning

AU - Yu, Xiangchun

AU - Yu, Zhezhou

AU - Wu, Lei

AU - Pang, Wei

AU - Lin, Chenghua

N1 - This work was supported by (1) the Science and Technology Developing Project of Jilin Province, China (Grant No. 20150204007GX) and (2) the Key Laboratory of Symbolic computation and Knowledge Engineering, Ministry of Education.

PY - 2019/3/8

Y1 - 2019/3/8

N2 - An important issue in statistical modeling is to determine the complexity of the model based on the scale of data so as to effectively mitigate the model’s overfitting problems without big data. We adopt a data-driven approach to automatically determine the number of components of the model. In order to better extract robust features, we propose a framework of data-driven two-layer structure visual dictionary learning (DTSVDL). It works by dividing the visual dictionary structure learning into two levels: the attribute layer and the detail layer. In the attribute layer, the attributes of the image dataset are learned, and these attributes are obtained by a data-driven Bayesian nonparametric model. Then, in the detail layer, the detailed information over attributes is further explored and refined, and the attributes are weighted by the number of effective observations associated with each attribute. Our proposed approach has three main advantages: (1) the two-layer structure makes our building visual dictionary be more expressive; (2) the number of components in the attribute layer can be determined automatically from the data; (3) the components are automatically determined based on the scale of visual words; therefore, our model can well mitigate the overfitting problem. In addition, by comparing with stacked autoencoders, stacked denoising autoencoders, LeNet-5, speeded-up robust features, and pretrained deep learning model ImageNet-VGG-F algorithms, we find that our approach achieves satisfactory image categorization results on two benchmark datasets. Specifically, higher categorization performance is achieved than by the classical approaches on 15 scene categories and action datasets. We conclude that the resulting DTSVDL possesses a good generality derived from attribute information as well as an excellent distinction derived from detailed information. In other words, the visual dictionary learned by our algorithm is more expressive and discriminatory.

AB - An important issue in statistical modeling is to determine the complexity of the model based on the scale of data so as to effectively mitigate the model’s overfitting problems without big data. We adopt a data-driven approach to automatically determine the number of components of the model. In order to better extract robust features, we propose a framework of data-driven two-layer structure visual dictionary learning (DTSVDL). It works by dividing the visual dictionary structure learning into two levels: the attribute layer and the detail layer. In the attribute layer, the attributes of the image dataset are learned, and these attributes are obtained by a data-driven Bayesian nonparametric model. Then, in the detail layer, the detailed information over attributes is further explored and refined, and the attributes are weighted by the number of effective observations associated with each attribute. Our proposed approach has three main advantages: (1) the two-layer structure makes our building visual dictionary be more expressive; (2) the number of components in the attribute layer can be determined automatically from the data; (3) the components are automatically determined based on the scale of visual words; therefore, our model can well mitigate the overfitting problem. In addition, by comparing with stacked autoencoders, stacked denoising autoencoders, LeNet-5, speeded-up robust features, and pretrained deep learning model ImageNet-VGG-F algorithms, we find that our approach achieves satisfactory image categorization results on two benchmark datasets. Specifically, higher categorization performance is achieved than by the classical approaches on 15 scene categories and action datasets. We conclude that the resulting DTSVDL possesses a good generality derived from attribute information as well as an excellent distinction derived from detailed information. In other words, the visual dictionary learned by our algorithm is more expressive and discriminatory.

KW - statistical modeling

KW - over-fitting

KW - visual dictionary

KW - Bayesian nonparametric model

KW - deep learning

KW - overfitting

KW - HIERARCHICAL MODEL

KW - WORDS

KW - BAG

KW - LATENT DIRICHLET ALLOCATION

KW - FEATURES

UR - http://www.mendeley.com/research/datadriven-twolayer-visual-dictionary-structure-learning

U2 - 10.1117/1.JEI.28.2.023006

DO - 10.1117/1.JEI.28.2.023006

M3 - Article

VL - 28

JO - Journal of Electronic Imaging

JF - Journal of Electronic Imaging

SN - 1017-9909

IS - 2

M1 - 023006

ER -