Fang, Kuangnan, Wang, Peng, Zhang, Xiaochen et al. · Figshare · 2020
In the application of high-dimensional data classification, several attempts have been made to achieve variable selection by replacing the ℓ2-penalty with other penalties for the support vector machine (SVM). However, these high-dimensional SVM methods usually do not take into account the special structure among covariates (features). In this article, we consider a classification problem, where the covariates are ordered in some meaningful way, and the number of covariates <i>p</i> can be much larger than the sample size <i>n</i>. We propose a structured sparse SVM to tackle this type of problems, which combines the non-convex penalty and cubic spline estimation procedure (i.e. penalizing second-order derivatives of the coefficients) to the SVM. From a theoretical point of view, the proposed method satisfies the local oracle property. Simulations show that the method works effectively both in feature selection and classification accuracy. A real application is conducted to illustrate the benefits of the method.
Cited by 0Open access
Meita Jayani, Samsuryadi Samsuryadi, Yunita Yunita · 2020
This research develops an application to recognize Lampung Script images using Structured Support Vector Machine (SSVM) method and measures the accuracy of the recognition results. The process of recognizing Lampung Script uses three processes, they are preprocessing by binarizing, cropping, skeletonizing, and resizing; feature extraction process using zone method; and recognition process using SSVM method. This research uses 1000 parent letter images as input. The average value of the recognition accuracy of Lampung Script is 22,56% for each pair of training data and test data. The amount of data used will affect the results of the recognition, the more training data used in the training process, the greater recognition succes accuracy will be obtained.
Cited by 0
Bach, Francis · SMARTech Repository (Georgia Institute of Technology) · 2021
Cited by 0
Thomas Hofmann, See Profile, Ioannis Tsochantaridis et al. · 2004
Learning general functional dependencies is one of the main goals in machine learning. Recent progress in kernel-based methods has focused on designing flexible and powerful in-put representations. This paper addresses the complementary issue of problems involv-ing complex outputs such as multiple depen-dent output variables and structured output spaces. We propose to generalize multiclass Support Vector Machine learning in a formu-lation that involves features extracted jointly from inputs and outputs. The resulting op-timization problem is solved efficiently by a cutting plane algorithm that exploits the sparseness and structural decomposition of the problem. We demonstrate the versatility and effectiveness of our method on problems ranging from supervised grammar learning and named-entity recognition, to taxonomic text classification and sequence alignment. 1.
Cited by 0
Yunpeng Li, Daniel P. Huttenlocher · 2008
We present a random field based model for stereo vision with explicit occlusion labeling in a probabilistic framework. The model employs non-parametric cost functions that can be learnt automatically using the structured support vector machine. The learning algorithm enables the training of models that are steered towards optimizing for a particular desired loss function, such as the metric used to evaluate the quality of the stereo labeling. Experimental results demonstrate that the performance of our method surpasses that of previous learning approaches and is comparable to the state-of-the-art for pixel-based stereo. Moreover, our method achieves good results even when trained on different image sets, in contrast with the common practice of hand tuning to specific benchmark images. In addition, we investigate the impact of graph structure on model performance. Our study shows that random field models with longer-range edges generally outperform the 4-connected grid and that this advantage is especially pronounced for noisy images.
Cited by 51
Hung-yi Lee, Yu-yu Chou, Yow-Bang Wang et al. · 2013
Supervised approaches can learn a spoken document summarizer generating high-quality summaries using a set of training examples matched to the domain of target documents. However, preparing a sufficient number of in-domain training examples is expensive. In this paper we propose an approach for unsupervised domain adaptation for spoken document summarization, so no in-domain training examples are needed. A summarizer is first learned from a set of out-of-domain training examples by a supervised summarization approach based on structured support vector machine, and this summarizer is used to generate a set of initial summaries for the target spoken documents. The target documents and their initial machine-generated summaries then serve as extra training examples for learning a new summarizer, which further updates the summaries of the target spoken documents. This process is continued iteratively to incrementally improve the summarizer for the target spoken documents. Moreover, extra approaches transforming the feature representations based on the data distribution in the target domain and augmenting the representations with an extra set of domain-specific features are also proposed. Encouraging results were obtained in summarizing Mandarin-English code-switching course lectures using training examples from Mandarin broadcast news.
Cited by 7
Hardoon, DR, Shawe-Taylor, J · UCL Discovery (University College London) · 2010
The tensor kernel has been used across the machine learning literature for a number of purposes and applications, due to its ability to incorporate samples from multiple sources into a joint kernel defined feature space. Despite these uses, there have been no attempts made towards investigating the resulting tensor weight in respect to the contribution of the individual tensor sources. Motivated by the increase in the current availability of Neuroscience data, specifically for two-source analyses, we propose a novel approach for decomposing the resulting tensor weight into its two components without accessing the feature space. We demonstrate our method and give experimental results on paired fMRI image-stimuli data. © The Author(s) 2009.
Cited by 0
Qinzhen Xu, Aiguo Song, Wenjiang Pei et al. · 2005
Support vector machines (SVMs) have gained wide acceptance because of the high generalization ability for a wide range of pattern recognition problems. We address problems associated with complex pattern recognition in this paper and present a tree-structured support vector machine (TSSVM) with confusion cross. A TSSVM is overall a binary tree, whose internal nodes are modular SVMs. Those two non-terminal nodes generated from the same parent node perform discounted confusion crossover. The presented approach is evaluated against other classifiers investigated lately. The performance of the proposed approach is demonstrated with some typical complex classification problems.
Cited by 5
Muhammad Febrian Rachmadi, Derry Alamsyah · SHILAP Revista de lepidopterología · 2017
Autonomous UAVs typically require systems capable of mapping (segmenting) the region of a colored image. Regional segmentation will be used to determine the existence of obstacles. Segmentation in an image initiated by grouping a number of pixels by color and proximity, with superpixel. Furthermore, the characteristics of each segmented region were extracted using Principal Component Analysis (PCA). Sequential segregation is a process of introducing a region that requires chain information obtained from the introduction before and after it called stochastic process. Solving this is done using Structured Support Vector Machines (SSVM). Meanwhile, to determine the weight in SSVM required a Conditional Random Field (CRF) model. The results show a good accuracy in the segmentation of the region that is 71%.
Cited by 1Open access
Jinyong Wu, Yong Zhao, Yule Yuan et al. · 2012
Since that the surveillance video is an unstructured media, it is not beneficial for the video intelligent retrieval and mining. An approach that is based on Gaussian mixture model and support vector machine has been put forward in this paper, which can make the video of surveillance scene structured. First, it constructs Gaussian background modeling to video scene, and isolates the motion object layer. Second, the visual perceptive information from moving object can be extracted by the angular point detecting method. Third, the multi-granularity perceptive feature of the object can be extracted by the object centroid-centred. Last, a 2-level SVM classifier should be build. By this classifier the semantics can be labeled to the moving objects, and then the structured description of the scenes can be obtained. The experimental results show that the presented method can avoid the interference caused by luminance changes and the motion of the leaves effectively. It is suitable for the video of surveillance scene in structured analysis application and can be a technical support for the intelligent retrieval and mining of video contents.
Cited by 0
Hung-yi Lee, Yu-yu Chou, Yow-Bang Wang et al. · 2012
Cited by 5
Bingbing Yu, Bo Wang, Yi Li et al. · Applied Intelligence · 2024
Cited by 1
Yingying Zhu, Xiaofeng Zhu, Minjeong Kim et al. · Lecture notes in computer science · 2016
Cited by 49Open access
Derry Alamsyah, Muhammad Febrian Rachmadi · SHILAP Revista de lepidopterología · 2018
Citra Kedalaman merupakan citra yang dapat digunakan untuk segmentasi objek, 3D Model dan visi pada robot. Penentuan nilai kedalaman dari sebuah citra warna memiliki tantangan tersendiri dan bidang aktif penelitian. Citra kedalaman dapat diestimasi dengan menggunakan model stokastik CRF. Selanjutnya model CRF digunakan sebagai model untuk pembobotan pada Model SSVM. Penggunaan CRF-SSVM untuk estimasi citra kedalaman didukung ekstraksi fitur domain frekuensi yaitu FFT. Hasil yang ditunjukan dengan menggunakan pendekatan ini sebesar 69.04%
Cited by 0Open access
Insuk Sohn, Sujong Kim, Jae‐Won Lee et al. · Model Assisted Statistics and Applications · 2017
NF-kB is a transcription factor (TF) which is an important mediator of immune responses. To understand human immune systems and related diseases, identifying novel NF-kB-responsive genes in the human genome is important. For this purpose, we developed a Discrete Kernel Structured Support Vector Mac hine (DSSVM) and applied this method to the promoters of 58 known NF-kB-target genes to find characteristic patterns of transcription factor binding sites (TFBSs) in their promoters. We searched these TFBS patterns against the promoters of 6440 genes previously identified in the human genome and identified 80 genes with the similar TFBS patterns in their promoters. Among them, 21 genes are well known NF-kB-target genes and 49 are putative NF-kB target genes associated with the immune system or inflammatory response. We confirmed that the selected genes function as NF-kB-target genes from a biological point of view. Our method of assigning functions to genes, based on distinctive patterns of TFBSs in their promoters, could be applicable to identification of, not only genes associated with a specific function, but genes with unique expression patterns.
Cited by 0
Lu Han, Sheng-syun Shen, Sz-Rung Shiang et al. · 2014
Cited by 4
Yonghui Ge, Siying Tu · Analytical Letters · 2022
AbstractAccurate detection of chilling injury in kiwifruit is challenging because the symptoms are mainly manifested in the interior. This work reports a method for detecting the chilling injury of 'Hongyang' kiwifruit to provide nondestructive discrimination. Kiwifruit samples with varying levels of chilling injury were analyzed by a hyperspectral structured-illumination reflectance imaging (SIRI) system. After demodulation, direct current (DC) and alternating current (AC) images with spatial frequencies of 30, 60, and 120 m−1 were obtained and labeled as F30, F60, and F120. Predictive models were developed to optimize the preprocessing and modeling methods. Prediction models established the results of DC and AC with different spatial frequencies and were compared. The autoscale-support vector machine (SVM) models were optimal for AC at different spatial frequencies, and the multiplicative scatter correction (MSC)-SVM model was optimal for DC. The combined features of F30, F60, and F120, as well as the spectral features of DC, had better accuracy for classifying chilling injury. The optimal model of hyperspectral SIRI system for detecting chilling injury was the F30 based on combined features, with calibration accuracy of 98.1% and prediction accuracy of 94.2%. This study has shown that structured illumination had higher accuracy than uniform illumination in predicting chilling injury. Further, this approach allows the identification of kiwifruit with chilling injury using a hyperspectral structured-illumination reflectance imaging system.Keywords: Chilling injuryhyperspectral structured illumination reflectance imaging (SIRI)kiwifruitpartial least squares discriminant analysis (PLS-DA)support vector machine (SVM) Disclosure statementThe authors declare that they have no known competing financial interests or personal relationships that influenced the work reported in this paper.Additional informationFundingThis work was supported by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).
Cited by 14
S. Manju, K. Helenprabha · Journal of Ambient Intelligence and Humanized Computing · 2022
Cited by 0Open accessRetracted
Шоучян, Кан, Микулович, А. В., Микулович, В. И. · Digital Library of the Belarusian State University (Belarusian State University) · 2012
An improved classification algorithm is proposed based on sphere structured multiclass support vector machine. Hyperspherical \nmulticlass support vector machine and new classification regulations are discussed in theory. Experimental research results show that \nbetter classification precision can be obtained by this method than ordinary hyperspherical multiclass support vector machine. = Предложен улучшенный алгоритм классификации на основе многоклассовой машины на опорных векторах (MOB) со \nструктурой в виде сферы. Приводится теоретическое описание гиперсферной многоклассовой MOB и нового правила классификации. Результаты экспериментальных исследований показывают, что данный метод позволяет достичь более высокой \nточности классификации по сравнению с обычной гиперсферной многоклассовой машиной на опорных векторах.
Cited by 0Open access
Yoonkyung Lee, Yuwon Kim, Sangjun Lee et al. · Biometrika · 2006
The support vector machine has been a popular choice of classification method for many applications in machine learning. While it often outperforms other methods in terms of classification accuracy, the implicit nature of its solution renders the support vector machine less attractive in providing insights into the relationship between covariates and classes. Use of structured kernels can remedy the drawback. Borrowing the flexible model-building idea of functional analysis of variance decomposition, we consider multicategory support vector machines with analysis of variance kernels in this paper. An additional penalty is imposed on the sum of weights of functional subspaces, which encourages a sparse representation of the solution. Incorporation of the additional penalty enhances the interpretability of a resulting classifier with often improved accuracy. The proposed method is demonstrated through simulation studies and an application to real data. Copyright 2006, Oxford University Press.
Cited by 35