Probability Estimates for Multi-class Classification by Pairwise Coupling

By Tingyao Wu, Chih-Jen Lin, Ruby Chiu-Hsing Weng. Journal of machine learning research, 2003

2,028 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/0d02083f5e9206ff88ed1e284da8665d4bee81ba

How researchers see this paper

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

Citations in the corpus

Data Mining with Neural Networks and Support Vector Machines Using the R/rminer Tool (2010, methodology): “The probabilistic SVM output is given by [19]: f(xi) = ∑m j=1 yjαjK(xj ,xi) + b p(i) = 1/(1 + exp(Af(xi) +B)) (3)”
Document Summarization Using Conditional Random Fields (2007, methodology): “Throughout this paper, we train support vector machines to do class probability estimation [Platt, 1999; Wu et al., 2004].”
An extensive experimental comparison of methods for multi-label learning (2012, methodology): “This group of methods consists of methods that use ensembles to make multi-label predictions and their base classi-ﬁers belong to either problem transformation or algorithm adaptation methods.”
A data-driven approach to predict the success of bank telemarketing (2014, methodology): “The probabilistic SVM output is given by [35]: f(xi) = ∑m j=1 yjαjK(xj,xi) + b and p(i) = 1/(1 + exp(Af(xi) + B)), where m is the number of support vectors, yi ∈ {−1, 1} is the output for a binary classification, b and αj are coefficients of the model, and A and B are determined by solving a regular”
An overview of ensemble methods for binary classifiers in multi-class problems: Experimental study on one-vs-one and one-vs-all schemes (2011, background): “Wu, Lin and Weng probability estimates by pairwise coupling approach (PE) [76] : PE is similar to PC, which also estimates the posterior probabilities (p) of each class starting from the pairwise probabilities.”
Label Distribution Learning (2013, methodology): “) is asso ciated with each label [33], [37], [15], [4].”
Automatic Classification for Pathological Prostate Images Based on Fractal Analysis (2009, methodology): “Two well-known extensions are one-against-one method [ 45 ], [46] and one-against-rest method [44], [47].”
Which Is the Best Multiclass SVM Method? An Empirical Study (2005, methodology): “A recent paper [12] proposed two new pairwise coupling schemes for estimation of class probabilities.”
Random-Forests-Based Network Intrusion Detection Systems (2008, methodology): “For instance, it has been applied to prediction [17], [25] and probability estimation [31].”
Short-term emotion assessment in a recall paradigm (2009, background): “For the second point, Platt (2000) proposes to model the probability of being in one of the two classes knowing the output value of the SVM by using a sigmoid fit, while Wu et al. (2004) proposes a solution to extend this idea to multiple classes.”
Text Classification Algorithms: A Survey (2019, methodology): “Voting After training all trees as forest, predictions are assigned based on voting [170] as follows:”
Sliced Wasserstein Kernels for Probability Distributions (2015, unknown): “[40] or a one versus all approach.”
Online Driver Distraction Detection Using Long Short-Term Memory (2011, methodology): “SVM-based classification of more than two classes was carried out by pairwise coupling according to [35].”
Fast and robust text detection in images and video frames (2005, methodology): “The probability of a candidate to be text is obtained in the SVM classification process as the method in [24].”
Forecasting cancellation rates for services booking revenue management using data mining (2010, methodology): “Platt (2000) and Wu et al. (2004) use logistic regression to link a data point’s distance to the hyperplane defined by ω∗ and b∗ with its class probability and thus making it possible to use SVM for probability estimation.”