Large-scale manifold learning

By Ameet Talwalkar, Sanjiv Kumar, Henry A. Rowley. 2008

236 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

SphereFace: Deep Hypersphere Embedding for Face Recognition (2017, background): “A-Softmax loss can be interpreted as constraining learned features to be discriminative on a hypersphere manifold, which intrinsically matches the prior that face images lie on a manifold [14, 5, 31].”
Multiview Spectral Embedding (2010, methodology): “In the future, we will consider how to utilize the sampling technique [26] in MSE to handle a large-scale data set.”
Laplacian Auto-Encoders: An explicit learning of nonlinear data manifold (2015, methodology): “, the Nyström method [36,37] or other landmarks based methods [38,39].”
Improved Bounds for the Nyström Method With Application to Kernel Classification (2011, unknown): “…to speed up kernel algorithms (Williams and Seeger, 2001; Drineas and Mahoney, 2005; Fowlkes et al., 2004; Kumar et al., 2009; Silva and Tenenbaum, 2003; Platt, 2004; Talwalkar et al., 2008; Zhang et al., 2008; Belabbas and Wolfe, 2009; Talwalkar and Rostamizadeh, 2010; Cortes et al., 2010).”
Matrix Coherence and the Nystrom Method (2010, methodology): “However, in practice, when working with real-world datasets, uniform sampling is more commonly used, e.g., Fowlkes et al. (2004); Platt (2004); Talwalkar et al. (2008); Williams and Seeger (2000), since diagonal sampling is more expensive and does not typically outperform uniform sampling (Kumar et…”
Joint Link Prediction and Attribute Inference Using a Social-Attribute Network (2014, methodology): “The attributes-to-weight mapping function can be learned using an approach similar to the one proposed by Backstrom and Leskovec [2011”
Sampling Methods for the Nyström Method (2012, unknown): “…method was first introduced to the machine learning community (Williams and Seeger, 2000) using uniform sampling without replacement, and this remains the sampling method most commonly used in practice (Talwalkar et al., 2008; Fowlkes et al., 2004; de Silva and Tenenbaum, 2003; Platt, 2004).”
Training Convolutional Networks with Noisy Labels (2014, background): “Nystrom methods (Talwalkar et al., 2008) can scale to large n, but do so by drastically sub-sampling first, resulting in a loss of fine structure within the problem.”
Parallel Spectral Clustering in Distributed Systems (2011, unknown): “[39] for manifold learning.”
A comprehensive survey of anomaly detection techniques for high dimensional big data (2020, background): “Density‑based Clustering‑based assumption that dimensionality reduction can be achieved by projecting the data onto a lower dimensional space where learning might be easier after such lower dimensional projection [65, 66].”
Inductive Hashing on Manifolds (2013, background): “In general, nonlinear manifold learning methods are more powerful than linear dimensionality reduction techniques, as they are able to more effectively preserve the local structure of the input data without assuming global linearity [26].”
A Graph based Analysis of Leak Localization in Urban Water Networks (2014, methodology): “The Nyström method is the widely adopted technique for approximating eigen-decomposition (Williams and Seeger, 2000; Fowlkes et al., 2004; Talwalkar et al., 2008).”
Revisiting the Nystrom Method for Improved Large-scale Machine Learning (2013, methodology): “nels below that have larger ˙parameter) are examples of relatively \nice" machine learning data sets that are similar to matrices where uniform sampling has been shown to perform well previously [56, 36, 37, 38]; and for these matrices our empirical results agree with these prior works. 3.3.3. ”
Efficient Label Propagation (2014, background): “Even though many approximation approaches have been proposed for efficient label propagation (Fergus et al., 2009; Kumar et al., 2009; Talwalkar et al., 2008; Yu & Yu, 2005; Zhu et al., 2003), one key advantage of our approach compared to them is that it guarantees the same label- Table 1.”