Which Is the Best Multiclass SVM Method? An Empirical Study

By Kaibo Duan, S. Sathiya Keerthi. International Workshop on Multiple Classifier Systems, 2005

709 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/93d648c333837a7090b54476e03e33d89a8a5d1d

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Multi-class Open Set Recognition Using Probability of Inclusion (2014, background): “However, this technique (like other estimators [26,53,56,17,28,6]) assumes that all scores must be from a known positive or known negative class.”
Feature selection via sensitivity analysis of SVM probabilistic outputs (2008, methodology): “The benefit of p(c|x) over f (x) in improving classification accuracy has been demonstrated on several numerical experiments (Platt 2000; Duan and Keerthi 2005).”
Learning Cumulatively to Become More Knowledgeable (2016, methodology): “As the decision score produced by SVM is not a probability distribution, several techniques have been proposed to convert a raw decision score to calibrated probabilistic outputs [2, 12, 17, 27, 41].”
Discriminative Semi-Supervised Feature Selection Via Manifold Regularization (2009, background): “For an overview on multiclass SVM, the readers can refer to [16], [25], where the performance of different implementations is compared.”
Probability Models for Open Set Recognition (2014, methodology): “Multiple heuristic techniques have been developed [8], [18] for converting multi-class SVM output to an estimated posterior probability.”
An introduction to continuous optimization for imaging (2016, background): “24 This can be slightly improved: see for instance Duan and Keerthi (2005).”
Probability Estimates for Multi-class Classification by Pairwise Coupling (2003, result): “Similar observations on waveform are also reported in (Duan and Keerthi, 2003), where the comparison is between δDV and δHT .”
Trajectory Optimization using Reinforcement Learning for Map Exploration (2008, background): “We have chosen to leverage a classifier known to generalize well in high-dimensional spaces, has few parameters to tune and is known to perform well on a variety of problems: a Support Vector Machine (Burges 1998; Duan and Keerthi 2005 ).”
Variational Bayesian Multinomial Probit Regression with Gaussian Process Priors (2006, methodology): “…class selection, MWVS (one-versus-one SVM with a maximum votes class selection strategy, PWCK (one-versus-one SVM with probabilistic outputs employing pairwise coupling; see Duan & Keerthi, 2005), for details, and PWCK (kernel logistic regression with pairwise coupling of binary outputs).”
Applied Predictive Modeling (2013, methodology): “Still more extensions have been developed for support vector machines, such as handling more than two classes (Hsu and Lin 2002; Duan and Keerthi 2005).”
Breaking the Closed World Assumption in Text Classification (2016, methodology): “As the decision score produced by SVM is not a probability distribution, several techniques have been proposed to convert a raw decision score to a calibrated probability output (Platt, 2000; Zadrozny and Elkan, 2002; Duan and Keerthi, 2005; Huang et al., 2006; Bravo et al., 2008).”