Kernel Logistic Regression and the Import Vector Machine

By Ji Zhu, Trevor J. Hastie. 2001

601 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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

Propensity score estimation: machine learning and classification methods as alternatives to logistic regression (2010, methodology): “The major difference between SVMs and logistic regression is that, although logistic regression attempts to explicitly model the probability (by means of the odds) of outcomes, SVMs attempt to directly find the best dividing hyperplane (or hyperplanes, in the case of more than two classes) regardles”
Discriminative Batch Mode Active Learning (2007, methodology): “An empirical study on UCI datasets shows that the proposed batch mode active learning method is more effective than some current state-of-the-art batch mode active learning algorithms.”
Robust Truncated Hinge Loss Support Vector Machines (2007, methodology): “…losses as surrogates of the 0–1 loss, for example the hinge loss of the SVM, the logistic loss in the penalized logistic regression (Lin et al. 2000; Zhu and Hastie 2005), and the exponential loss function used in AdaBoost (Friedman, Hastie and Tibshirani 2000 Despite its wide success, however,…”
Large-scale sparse logistic regression (2009, background): “ing document classification [4, 5], computer vision [9], natural language processing [14, 22], and bioinformatics [2, 20, 29, 34, 35].”
Face Liveness Detection from a Single Image with Sparse Low Rank Bilinear Discriminative Model (2010, background): “This characteristic is shared by many sparse nonlinear discriminative models in literatures, such as probabilistic Support Vector Machine (pSVM, [22]), Relevance Vector Machine (RVM, [26]) and Import Vector Machine (IVM, [29]).”
Exponentiated Gradient Algorithms for Conditional Random Fields and Max-Margin Markov Networks (2008, methodology): “Earlier work (Jaakkola and Haussler, 1999; Keerthi et al., 2005; Zhu and Hastie, 2001) treated the logistic regression model, a simpler version of a CRF.”
Approximation Methods for Gaussian Process Regression (2007, methodology): “The question of how to choose the inducing inputs still arises; Lin et al. (2000) select these using a clustering method, while Zhu and Hastie (2002) propose a forward selection strategy.”
Forecasting cancellation rates for services booking revenue management using data mining (2010, methodology): “Recently, kernel based methods, notably Support Vector Machine (SVM) (Platt 2000) and Kernel Logistic Regression (KLR) (Wahba 1999, Zhu and Hastie 2005), have been developed for class probability estimation and obtained promising results.”
Trading Interpretability for Accuracy: Oblique Treed Sparse Additive Models (2015, background): “It is well-recognized that, despite the dramatic evolution of machine learning approaches, such as kernel machines [41, 45], boosting [13], random forests [3], and deep decision trees, are still preferred in such applications as marketing, medical analytics, and science, for which understanding…”
Training a Support Vector Machine in the Primal (2007, background): “Some other losses have been proposed to approximate the SVM hinge loss (Lee and Mangasarian, 2001b; Zhang et al., 2003; Zhu and Hastie, 2005).”
Decentralized Online Learning With Kernels (2017, methodology): “Gaussian Mixture Model Following [24], [31], we generate a data set from Gaussian mixture models, which consists N = 5000 feature-label pairs for training and 2500 for testing.”