Memory Efficient Kernel Approximation

By Si Si, Cho-Jui Hsieh, Inderjit S. Dhillon. 2014

155 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.

Citations in the corpus

FastMMD: Ensemble of Circular Discrepancy for Efficient Two-Sample Test (2014, background): “Many efforts have been made to speedup the establishment of the kernel information and accelerate the implementation of kernel techniques (Smola & Schölkopf, 2000; Williams & Seeger, 2000; Fine & K.Scheinberg, 2001; Si et al., 2014).”
Nyströmformer: A Nystöm-based Algorithm for Approximating Self-Attention (2021, methodology): “…Nystr ¨ om with pseudo landmarks, prototype method (Wang and Zhang 2013; Wang, Zhang, and Zhang 2016), fast-Nys (Si, Hsieh, and Dhillon 2016), and MEKA (Si, Hsieh, and Dhillon 2017), ensemble Nystr¨om (Kumar, Mohri, and Talwalkar 2009) have been proposed for spe-ciﬁc improvements over the basic…”
Combiner: Full Attention Transformer with Sparse Computation Cost (2021, background): “However such random-feature based approaches [29] require many more bases to better approximate the exponential kernel [26], and empirically we found it produces inferior results in some sequence modeling tasks, such as density estimation.”
Less is More: Nyström Computational Regularization (2015, background): “In particular, we prove that these approaches can achieve optimal learning bounds, provided the subsampling level is suitably chosen.”
Approximate Log-Hilbert-Schmidt Distances between Covariance Operators for Image Classification (2016, background): “There exists a large literature on large-scale kernel approximation, which focuses on approximating either the feature maps or the kernel matrices [25, 34, 27], but not on the covariance operators.”
Likelihood approximation with hierarchical matrices for large spatial datasets (2017, methodology): “In the last few years, there has been great interest in numerical methods for representing and approximating large covariance matrices in the applied mathematics community [54, 12, 37, 59, 50, 3, 2, 61, 4].”