Semi-supervised dimensionality reduction in image feature space

Cheng, Hao; Hua, Kien A.; Vu, Khanh; Liu, Danzhou

doi:10.1145/1363686.1363966

Cited by 5 publications

(6 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Through conducting many experiments on five popular multi-label datasets, the novel MSDA algorithm was compared with other dimensionality reduction methods: ORI (original dataset), PCA [2], LTSA [2], MDDM [9], and MLSI [8]. In particular, ORI as a baseline deals with the original dataset directly.…”

Section: Methodsmentioning

confidence: 99%

“…Parameters used in experiments were set as follows: LPP, LTSA: k=10; MLSI was the same as that in Ref. [8]: β=0.5; MSDA: the default labeled samples ratio was 30%, α=0. 8, β=0.85, the number of neighbors of each sample was 10; and multi-label k nearest neighbor (MLKNN): k=10 [9].…”

Section: Environment Settingsmentioning

confidence: 99%

“…SONG et al [7] put forward a unified framework for semi-supervised dimensionality reduction, under which PCA, LDA, MMC (maximum margin criterion), LPP and their corresponding kernel versions were all covered. CHENG et al [8] developed a semi-supervised dimensionality reduction method to balance the relevant feedback information in the image feature space. The above methods all aim at single-label problem, but luckily those methods have provided some useful minds for multi-label dimensionality reduction, which would be clearly shown in the following parts.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Multi-label dimensionality reduction based on semi-supervised discriminant analysis

Guo

et al. 2010

J. Cent. South Univ. Technol.

View full text Add to dashboard Cite

Multi-label data with high dimensionality often occurs, which will produce large time and energy overheads when directly used in classification tasks. To solve this problem, a novel algorithm called multi-label dimensionality reduction via semi-supervised discriminant analysis (MSDA) was proposed. It was expected to derive an objective discriminant function as smooth as possible on the data manifold by multi-label learning and semi-supervised learning. By virtue of the latent imformation, which was provided by the graph weighted matrix of sample attributes and the similarity correlation matrix of partial sample labels, MSDA readily made the separability between different classes achieve maximization and estimated the intrinsic geometric structure in the lower manifold space by employing unlabeled data. Extensive experimental results on several real multi-label datasets show that after dimensionality reduction using MSDA, the average classification accuracy is about 9.71% higher than that of other algorithms, and several evaluation metrices like Hamming-loss are also superior to those of other dimensionality reduction methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Environment Settingsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi-label dimensionality reduction based on semi-supervised discriminant analysis

Guo

et al. 2010

J. Cent. South Univ. Technol.

View full text Add to dashboard Cite

show abstract

“…Dimensionality Reduction [3], Data Co-Reduction [5] and Hashing methods [4,9], to accelerate the search process of the high-dimensional image vectors, a.k.a. descriptor vectors.…”

Section: Introductionmentioning

confidence: 99%

Parallel similarity search based on the dimensions value cardinalities of image descriptor vectors

Rafailidis

Manolopoulos

2015

Proceedings of the 30th Annual ACM Symposium on Applied Computing

View full text Add to dashboard Cite

In this paper, we propose a parallel similarity search strategy based on the dimensions value cardinalities, an inherit characteristic of image descriptor vectors. Our strategy has low preprocessing requirements by dividing the computational cost of the preprocessing steps into several machines and locating the descriptors with similar dimensions value cardinalities logically close. Additionally, an efficient parallel query processing algorithm is proposed, where the dimensions of image descriptors are prioritized in the searching strategy, assuming that dimensions of high value cardinalities have more discriminative power than the dimensions of low ones. In our experiments with publicly available datasets of 80 million and 1 billion images, we show that the proposed method outperforms state-of-the-art parallel similarity search strategies, in terms of preprocessing cost, search time and accuracy. Finally, we made our source code publicly available.

show abstract

“…Therefore, several approximate similarity search strategies were proposed, performing an approximation of the visual nearest neighbors of sequential search, by highly reducing the computational cost. For example, approximate similarity search strategies are the Dimensionality Reduction [Cheng et al 2008;Liang et al 2010; Wang and Binbin 2011], Data Co-Reduction [Huang et al 2011], Vantage Indexing [Bozkaya and Ozsoyoglu 1999;Fu et al 2000;Van Leuken and Veltkamp 2011] and Hashing methods [Gionis et al 1999;Jegou et al 2011;Heo et al 2012;Liu et al 2014]. Recently, the MSIDX method [Tiakas et al 2013] exploited a new key factor of the image descriptor vectors, namely the dimensions value cardinalities.…”

Section: Introductionmentioning

confidence: 99%

Similarity Search over the Cloud Based on Image Descriptors' Dimensions Value Cardinalities

Antaris

Rafailidis

2015

ACM Trans. Multimedia Comput. Commun. Appl.

View full text Add to dashboard Cite

In recognition that in modern applications billions of images are stored into distributed databases in different logical or physical locations, we propose a similarity search strategy over the cloud based on the dimensions value cardinalities of image descriptors. Our strategy has low preprocessing requirements by dividing the computational cost of the preprocessing steps into several nodes over the cloud and locating the descriptors with similar dimensions value cardinalities logically close. New images are inserted into the distributed databases over the cloud efficiently, by supporting dynamical update in real-time. The proposed insertion algorithm has low computational complexity, depending exclusively on the dimensionality of descriptors and a small subset of descriptors with similar dimensions value cardinalities. Finally, an efficient query processing algorithm is proposed, where the dimensions of image descriptors are prioritized in the searching strategy, assuming that dimensions of high value cardinalities have more discriminative power than the dimensions of low ones. The computation effort of the query processing algorithm is divided into several nodes over the cloud infrastructure. In our experiments with seven publicly available datasets of image descriptors, we show that the proposed similarity search strategy outperforms competitive methods of single node, parallel and cloud-based architectures, in terms of preprocessing cost, search time and accuracy. ACM Reference Format:Stefanos Antaris and Dimitrios Rafailidis. 2015. Similarity search over the cloud based on image descriptors' dimensions value cardinalities. ACM Trans. Multimedia Comput. Commun.

show abstract

Semi-supervised dimensionality reduction in image feature space

Cited by 5 publications

References 14 publications

Multi-label dimensionality reduction based on semi-supervised discriminant analysis

Multi-label dimensionality reduction based on semi-supervised discriminant analysis

Parallel similarity search based on the dimensions value cardinalities of image descriptor vectors

Similarity Search over the Cloud Based on Image Descriptors' Dimensions Value Cardinalities

Contact Info

Product

Resources

About