Less is More: Nystr\"om Computational Regularization

128 citations recorded by the source corpus.

How researchers see this paper

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

Learning Theory for Distribution Regression (2014, methodology): “ distributions to a reproducing kernel Hilbert space, which has a simple analytical solution, as well as theoretically sound, eﬃcient methods for approximation (Zhang et al., 2014; Yang et al., 2015; Rudi et al., 2015). We derived explicit bounds on the excess risk as a function of the number of sam”
Hybrid Random Features (2021, methodology): “Several other techniques such as Nystr¨om method (Yang et al., 2012; Williams & Seeger, 2000; Rudi et al., 2015) were proposed to construct data-dependent feature representations.”
Quantum machine learning: a classical perspective (2017, methodology): “tically provides an improvement in computational complexity. It has been recently shown that it is possible to obtain equivalent generalisation performance to classical methods also in these settings [RCR15]. For all these methods, training times can be typically reduced from the O(N3) of standard a”
Harder, Better, Faster, Stronger Convergence Rates for Least-Squares Regression (2016, background): “pproximation in large scale learning, as described by Bottou and Bousquet (2008). In a similar ﬂavor, it also appears that regularization can be accomplished through early stopping (Yao et al., 2007; Rudi et al., 2015), highlighting this interplay between computation and statistics. To characterize ”