A novel formulation of orthogonal polynomial kernel functions for SVM classifiers: The Gegenbauer family

Padierna, Luis Carlos; Carpío, Martín; Rojas-Domínguez, Alfonso; Soberanes, Héctor José Puga; Fraire, Héctor

doi:10.1016/j.patcog.2018.07.010

Cited by 62 publications

(12 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consider the nonlinear transformation Φ : R m ⟶ ℋ that allows to project the input vectors from the original coordinate space x to a transformed space Φ( x ) ∈ ℋ , provided with scalar product. The scalar product between two given vectors in the transformed space enunciated in terms of a function of similarity in the original space Φ( u ) · Φ( v )= K ( u , v ) is known as the kernel function [30, 31].…”

Section: Classification Modelsmentioning

confidence: 99%

Predictive Models for the Medical Diagnosis of Dengue: A Case Study in Paraguay

Mello-Román

Román

Gómez-Guerrero

et al. 2019

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

Early diagnosis of dengue continues to be a concern for public health in countries with a high incidence of this disease. In this work, we compared two machine learning techniques: artificial neural networks (ANN) and support vector machines (SVM) as assistance tools for medical diagnosis. The performance of classification models was evaluated in a real dataset of patients with a previous diagnosis of dengue extracted from the public health system of Paraguay during the period 2012–2016. The ANN multilayer perceptron achieved better results with an average of 96% accuracy, 96% sensitivity, and 97% specificity, with low variation in thirty different partitions of the dataset. In comparison, SVM polynomial obtained results above 90% for accuracy, sensitivity, and specificity.

show abstract

Section: Classification Modelsmentioning

confidence: 99%

Predictive Models for the Medical Diagnosis of Dengue: A Case Study in Paraguay

Mello-Román

Román

Gómez-Guerrero

et al. 2019

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

show abstract

“…The feature extraction step is made in order to find the most valuable features and to reduce the amount of data that describes the object. There are many efficient feature extractors used for detection of pedestrians, starting with the basic handcrafted features like histograms of oriented gradients (HOG) [34], local binary patterns (LBP) [35], shape context [36], 1D/2D Haar descriptors [37], to plenty of their modifications [17,19,30,36,38]. Recently, several efficient variants of the HOG were proposed: integral channel features (ICF), for which the HOG descriptors are used together with luminance and UV chrominance components (LUV) [39], the ACF [40] combining HOG channel feature with the normalized gradient magnitude and LUV color channels, and the Checkerboards [41], which are modifications of the ICF.…”

Section: Features Extractionmentioning

confidence: 99%

“…In the experiments, we decided to use three various baseline detectors, namely: histogram of oriented gradients (HOG) with the support vector machine (SVM) classifier, the aggregated channel feature (ACF) detector, and the deep convolutional neural network (CNN). The first two are commonly used in the standard real-time applications for the detection of pedestrians in IR images [9,10,[16][17][18][19][20]. Recently, we have observed a rapid and promising development of various classifiers and detectors based on CNNs.…”

Section: Introductionmentioning

confidence: 99%

Tuning of Classifiers to Speed-Up Detection of Pedestrians in Infrared Images

Piniarski

Pawlowski

Dąbrowski

2020

Sensors

View full text Add to dashboard Cite

This paper presents an experimental evaluation of real-time pedestrian detection algorithms and their tuning using the proposed universal performance index. With this index, the precise choice of various parameters is possible. Moreover, we determined the best resolution of the analysis window, which is much lower than the initial window. By such means, we can speed-up the processing (i.e., reduce the classification time by 74%). There are cases in which we increased both the processing speed and the classification accuracy. We made experiments with various baseline detectors and datasets in order to confirm versatility of the proposed ideas. The analyzed classifiers are those typically applied to detection of pedestrians, namely: aggregated channel feature (ACF), deep convolutional neural network (CNN), and support vector machine (SVM). We used a suite of five precisely chosen night (and day) IR vision datasets.

show abstract

“…Moreover, kernel selection is becoming more challenging because the number of valid kernels continues to grow as new kernel families are proposed. These families include: wavelet kernels [31], non-parametric kernels [32], and orthogonal polynomial kernels [33] [34] [35] [36]. Kernels in TABLE I have proved to be valid kernels since they satisfy the necessary and sufficient conditions established in Mercer's theorem [37].…”

Section: A Kernel Functions For Support Vector Machinesmentioning

confidence: 99%

“…[38]). This approach has been followed in several relevant works [14] [36], and has introduced the concept of a Multiple Kernel, denoted:…”

Section: A Kernel Functions For Support Vector Machinesmentioning

confidence: 99%

Biomedical Classification Problems Automatically Solved by Computational Intelligence Methods

2020

View full text Add to dashboard Cite

Biomedical classification problems are of great interest to both medical practitioners and computer scientists. Due to the harmful consequences of a wrong decision in this ambit, computational methods must be carefully designed to provide a reliable tool for helping physicians to obtain accurate predictions on unseen cases. Computational Intelligence (CI) provides robust models to perform optimization, classification and regression tasks. These models have been previously designed, mainly based on the expertise of computer scientists, to solve a vast number of biomedical problems. As the number of both CI algorithms and biomedical problems continues to grow, selecting the right method to solve a given problem becomes more challenging. To deal with this complexity, a systematic methodology for selecting a suitable model for a given classification problem is required. In this work, we review the more promising classification and optimization algorithms and reformulate them into a synergic framework to automatically design and optimize pattern classifiers. Our proposal, including state-of-the-art evolutionary algorithms and support vector machines, is tested on a variety of biomedical problems. Experimental results on benchmark datasets allow us to conclude that the automatically designed classifiers reach higher or equal performance than those designed by computer specialists.

show abstract

A novel formulation of orthogonal polynomial kernel functions for SVM classifiers: The Gegenbauer family

Cited by 62 publications

References 18 publications

Predictive Models for the Medical Diagnosis of Dengue: A Case Study in Paraguay

Predictive Models for the Medical Diagnosis of Dengue: A Case Study in Paraguay

Tuning of Classifiers to Speed-Up Detection of Pedestrians in Infrared Images

Biomedical Classification Problems Automatically Solved by Computational Intelligence Methods

Contact Info

Product

Resources

About