On Kernel-Target Alignment

By Nello Cristianini, John Shawe-Taylor, Andre´ Elisseeﬀ, Jaz S. Kandola. Neural Information Processing Systems, 2001

1,241 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/36aa0d0936b2cf128c646c36a1981807b5a27aaf

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 Statistics and Probability

Citations in the corpus

Learning Kernels with Random Features (2016, neutral): “The latter approaches require an eigendecomposition of the Gram matrix or costly optimization problems (e.g. quadratic or semidefinite programs) [10] [14], but these models have a variety of generalization guarantees [1] [8] [10] [18] [19].”
GeneMANIA: a real-time multiple association network integration algorithm for predicting gene function (2008, background): “, αd], is customized to a particular prediction task by choosing it to maximize a form of kernel-target alignment [39] between the composite network and a 'target' network constructed from the class label vector y.”
First-Person Activity Recognition: What Are They Doing to Me? (2013, methodology): “We first introduce kernel target alignment [15] that measures the angle between two Gram matrices, and present that it can be used to evaluate structure kernels for our activity recognition.”
Hyperparameter Learning for Graph Based Semi-supervised Learning Algorithms (2006, background): “also opened window to learning the kernel by maximizing the kernel alignment (Cristianini et al., 2001) or low rank mapping to feature space (Kwok & Tsang, 2003), with respect to those hyperparameters σ, a, t.”
Similarity of Neural Network Representations Revisited (2019, background): “This normalized index is known as centered kernel alignment (Cortes et al., 2012; Cristianini et al., 2002): CKA(K,L) = HSIC(K,L)√ HSIC(K,K)HSIC(L,L) .”
Domain Adaptation via Transfer Component Analysis (2009, background): “Mathematically, this leads to the following semi-deﬁnite program (SDP): max K ≽ 0 tr ( HKHK yy ) subject to constraints on K. ( In particular, (1) reduces to Maximum Variance Unfolding (MVU) [26] when no side information is given (i.e., K yy = I ).”
Supervised feature selection via dependence estimation (2007, methodology): “Likewise, one could use Kernel Target Alignment (KTA) [11] to test directly whether there exists any correlation between data and labels.”
Learning The Discriminative Power-Invariance Trade-Off (2007, background): “Of particular interest are [3, 4, 35, 42] as each learns the optimal kernel for classiﬁcation as a linear combination of base kernels with positive weights while enforcing sparsity.”
Multiple Kernel Learning Algorithms (2011, neutral): “This corresponds to a function that assigns weights to kernel functions: 4.1 Combining Kernels using Kernel Alignment Cristianini et al. (2002) deﬁne a notion of similarity between two kernels called kernel alignment .”
Text Classification using String Kernels (2002, methodology): “This section reviews the main ideas behind Support Vector Machines (SVMs) and kernel functions.”
Kernel methods in machine learning (2007, neutral): “ where h·,·iF is the Frobenius dot product. If the two classes have equal size, then up to a scaling factor involving kKk 2 and n, this equals the kernel-target alignment deﬁned by Cristianini et al. [38]. 2.2. Positive deﬁnite kernels. We have required that a kernel satisfy (3), that is, correspond”
Gaussian Processes for Machine Learning (Adaptive Computation and Machine Learning) (2005, neutral): “For example Cristianini et al. [2002] define the alignment between a Gram matrix K and the corresponding +1/− 1 vector of alignment targets y as A(K,y) = y>Ky n‖K‖F , (5.”
l p -Norm Multiple Kernel Learning (2011, background): “Kernel target alignment (Cristianini et al., 2002; Cortes et al., 2010b) aims at learning the entries of a kernel matrix by using the outer product of the label vector as the groundtruth.”
High-Dimensional Feature Selection by Feature-Wise Kernelized Lasso (2012, background): “…selection (Song et al. , 2012), quadratic programming feature selection (QPFS) (Rodriguez-Lujan et al. , 2010), kernel target alignment (KTA) (Shawe-Taylor and Kandola, 2002; Cortes et al. , 2012), Hilbert-Schmidt Feature Selection (HSFS) (Masaeli et al. , 2010), and sparse additive models…”
A Binary Classification Framework for Two-Stage Multiple Kernel Learning (2012, methodology): “Previous two-stage kernel learning approaches (Cristianini et al., 2001; Cortes et al., 2010b) learn a non-negative linear combination of base kernels that maximizes the alignment with the target kernel K(xi,xj) = yiyj on the training set.”
Discriminative Semi-Supervised Feature Selection Via Manifold Regularization (2009, background): “A number of performance criteria have been proposed for feature selection, including mutual information [19], [31], maximum margin [22], [52], kernel alignment [15], [37], graph-spectrum based measures [53], [66], construction errors in neural network [35], [42], worst case probability [4], [61], an”