Multiple kernel learning, conic duality, and the SMO algorithm

By Francis R. Bach, Gert R. G. Lanckriet, Michael I. Jordan. 2004

1,667 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Mechanics, Numerical Analysis, Statistical Learning Theory, Mathematical Optimization

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 Computational Mechanics

In Numerical Analysis

In Statistical Learning Theory

In Mathematical Optimization

Citations in the corpus

Kernel Continual Learning (2021, methodology): “To address this, we further introduce random Fourier features to learn data-driven kernels, which have previously demonstrated success in regular learning tasks (Bach et al., 2004; Sinha & Duchi, 2016; Carratino et al., 2018; Zhen et al., 2020).”
Testing for Homogeneity with Kernel Fisher Discriminant Analysis (2007, methodology): “s separation) is likely to yield greater power. As future work, we plan to investigate, both theoretically and practically, the use of (single and multiple) kernel learning procedures as developed by Bach et al. (2004) for maximizing the expected power of our test statistics in speciﬁc applications.”
A Survey on Multi-view Learning (2013, methodology): “Keywords: Multi-view Learning, Survey, Machine Learning”
Joint covariate selection and joint subspace selection for multiple classification problems (2010, methodology): “This approach is a particular case of a general methodology in which block norms are used to define groupings of variables in regression and classification problems (Bach et al. 2004; Yuan and Lin 2006; Park and Hastie 2006; Meier et al. 2008; Kim et al. 2006; Zhao et al. 2008).”
Optimization with Sparsity-Inducing Penalties (2011, neutral): “26) using sparsityinducing norms (see [7, 13, 110] for more details).”
Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification (2015, methodology): “The fusion weights can be manually set or automatically estimated by multiple kernel learning (MKL) [2].”
Audio-Visual Instance Discrimination with Cross-Modal Agreement (2020, background): “Classical approaches can be broadly categorized in co-training procedures [6, 7, 30, 43, 50, 67, 84] that maximize the mutual agreement between views, multiple kernel learning procedures [5, 40, 44] which use kernels to model different views, and subspace learning procedures [15, 68] which seek to f”
More efficiency in multiple kernel learning (2007, methodology): “Connections with 1-norm Block Regularization Formulation of MKL The multiple kernel learning formulation introduced by Bach et al. (2004) and further developed by Sonnenburg et al. (2006) consists in solving the following optimization problem: 21 minwk,b,.i ( JwkJ)+ C .i 2 k i with yi…”
Inferring latent task structure for Multitask Learning by Multiple Kernel Learning (2010, background): “Lanckriet et al. considered conic combinations of kernel matrices for classification [ 16 ], leading to a convex quadratically constrained quadratic program.”
A Survey of Modern Questions and Challenges in Feature Extraction (2015, background): “…families K have been studied (Bach et al., 2004; Weinberger et al., 2004; Cortes et al., 2010b, 2013; Lanckriet et al., 2004; Cortes et al., 2010b; Bach et al., 2004), but the most widely used family is that of non-negative linear combinations of base kernels K = { ∑p i=1 µiKi : µ ∈ Rp, ‖µ‖q ≤…”
Neighborhood repulsed metric learning for kinship verification (2012, methodology): “is the multiple kernel learning method (MKL) [3] where”
Generalized Composite Kernel Framework for Hyperspectral Image Classification (2013, background): “On the other hand, recent studies have shown that composite kernels [17] and multiple kernels [18]–[22] can provide enhanced classification accuracy by incorporating the spatial information in addition to the spectral information.”
Multiple Kernel Learning from Noisy Labels by Stochastic Programming (2012, methodology): “Many algorithms have been developed for maxmargin MKL by formulating the problem into SemiDefinite Programming (Lanckriet et al., 2004), Second Order Cone Programming (Bach et al., 2004), and Semi-Infinite Linear Programming (Sonnenburg et al., 2006).”
TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 1 ModDrop (2018, positive): “Nataranjan et al. [45] employed multiple strategies, including MKL-based combinations of features, Bayesian model combination, and weighted average fusion of scores from multiple systems.”
Salient object detection via bootstrap learning (2015, methodology): “The MKB algorithm is a boosted Multiple Kernel Learning (MKL) method [4], which combines several SVMs of different kernels.”