Classification Performance Comparison between RVM and SVM

Xiangmin, Xu; Mao, Yun-feng; Jia-ni, Xiong; Feng-le, Zhou

doi:10.1109/iwasid.2007.373728

Cited by 22 publications

(9 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [8], it was also shown that RVMs have a much lower minimum descriptor length with up to 28% reduction in the the number of decision vectors required across all features used in the experiment. To further support this, experiments by [40] show that RVMs are less susceptible to over fitting when compared to SVM. This is a desirable characteristic as we want our pain recognition model to generalize well to real world data.…”

Section: Pain Intensity Estimationmentioning

confidence: 92%

Cumulative attributes for pain intensity estimation

Egede

Valstar

2017

Proceedings of the 19th ACM International Conference on Multimodal Interaction

View full text Add to dashboard Cite

Pain estimation from face video is a hard problem in automatic behaviour understanding. One major obstacle is the difficulty of collecting sufficient amounts of data, with balanced amounts of data for all pain intensity levels. To overcome this, we propose to adopt Cumulative Attributes, which assume that attributes for high pain levels with few examples are a superset of all attributes of lower pain levels. Experimental results show a consistent relative performance increase in the order of 20% regardless of features used. Our final system significantly outperforms the state of the art on the UNBC McMaster Shoulder Pain database by using cumulative attributes with Relevance Vector Regression on a combination of features, including appearance, geometric, and deep learned features.

show abstract

Section: Pain Intensity Estimationmentioning

confidence: 92%

Cumulative attributes for pain intensity estimation

Egede

Valstar

2017

Proceedings of the 19th ACM International Conference on Multimodal Interaction

View full text Add to dashboard Cite

show abstract

“…Note that w MP , the maximum posterior of w, is obtained by Laplace's Method in [24], which approximates a Normal distribution with the mean value w 0 to the posterior of w. [4,15,30] concluded that RVM is better than SVM in the fields of classification and regression. They also showed that the conduction speed of RVM is faster than SVM.…”

Section: (4)mentioning

confidence: 99%

“…All the variables in X −1,j and X 1,j have the class labels −1 (Majority) and 1 (Minority), respectively. We set five kinds of sizes, (n p , n n ) = (30,30), (15,30), (12,30), (6,30), and (3, 30) to illustrate the performance of different algorithms in different-sized data. b = 1, 2, 2.5, 5, 10 for these five cases and a larger b indicates a more severely imbalanced dataset.…”

Section: Numeric Studies a Simulation Data Studiesmentioning

confidence: 99%

Fully Bayesian Analysis of Relevance Vector Machine Classification With Probit Link Function for Imbalanced Data Problem

et al. 2021

View full text Add to dashboard Cite

The original RVM classification model uses the logistic link function to build the likelihood function making the model hard to be conducted since the posterior of the weight parameter has no closed-form solution. This paper proposes the probit link function approach instead of the logistic one for the likelihood function in the RVM classification model, namely PRVM (RVM with the probit link function). We show that the posterior of the weight parameter in PRVM follows the Multivariate Normal distribution and achieves a closed-form solution. A latent variable is needed in our algorithms to simplify the Bayesian computation greatly, and its conditional posterior follows a truncated Normal distribution. Compared with the original RVM classification model, our proposed one is a Fully Bayesian approach, and it has a more efficient computation process. For the prior structure, we first consider the Normal-Gamma independent prior to propose a Generic Bayesian PRVM algorithm. Furthermore, the Fully Bayesian PRVM algorithm with a hierarchical hyperprior structure is proposed, which improves the classification performance, especially in the imbalanced data problem.

show abstract

“…Regarding this, in this paper an authentication approach using Relevance Vector Machine (RVM) [12] proposed. RVM is also like SVM -a sparse linearly parameterized model -but RVM requires fewer relevant vectors [12], [13]; hence, it increases the speed of the authentication process. This paper presents a method using face and fingerprint traits and feature level fusion for the aims of finding effective ways to fuse at feature level of different characteristics, and constructing templates from the combined features.…”

Section: Introductionmentioning

confidence: 99%

Person Authentication using Relevance Vector Machine (RVM) for Face and Fingerprint

Tran

Le²

2015

IJMECS

View full text Add to dashboard Cite

Abstract-Multimodal biometric systems have proven more efficient in personal verification or identification than single biometric ones, so it is also a focus of this paper. Particularly, in the paper, the authors present a multimodal biometric system in which features from face and fingerprint images are extracted using Zernike Moment (ZM), the personal authentication is done using Relevance Vector Machine (RVM) and feature-level fusion technique. The proposed system has proven its remarkable ability to overcome the limitations of unimodal biometric systems and to tolerate local variations in the face or fingerprint image of an individual. Also, the achieved experimental results have demonstrated that using RVM can assure a higher level of forge resistance and enables faster authentication than the state-of-the-art technique , namely the support vector machine (SVM).

show abstract

Classification Performance Comparison between RVM and SVM

Cited by 22 publications

References 1 publication

Cumulative attributes for pain intensity estimation

Cumulative attributes for pain intensity estimation

Fully Bayesian Analysis of Relevance Vector Machine Classification With Probit Link Function for Imbalanced Data Problem

Person Authentication using Relevance Vector Machine (RVM) for Face and Fingerprint

Contact Info

Product

Resources

About