An extension of slow feature analysis for nonlinear blind source separation

By Henning Sprekeler, Tiziano Zito, Laurenz Wiskott. 2014

76 citations recorded by the source corpus.

Subfields: Analytical Chemistry, Computer Vision and Pattern Recognition, Signal Processing

How researchers see this paper

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

In Analytical Chemistry

1. Aapo Hyvärinen
2. Yash Sharma
3. Hermanni Hälvä
4. Wieland Brendel
5. Hiroaki Sasaki
6. Richard E. Turner
7. Hiroshi Morioka
8. J. Urgen Schmidhuber

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Trevor Darrell
3. Dacheng Tao
4. Ming-Hsuan Yang
5. Xiaogang Wang
6. Andrea Vedaldi
7. Andrew Zisserman
8. Yoshua Bengio

In Signal Processing

1. J. Urgen Schmidhuber
2. Aapo Hyvärinen
3. Yash Sharma
4. Hermanni Hälvä
5. Wieland Brendel
6. Bernhard Schölkopf
7. Matthias Bethge
8. Richard E. Turner

Citations in the corpus

• Hidden Markov Nonlinear ICA: Unsupervised Learning from Nonstationary Time Series (2020, methodology): “For linear ICA this approach has been shown to yield identifiable models (Belouchrani et al., 1997; Tong et al., 1991), and extensions to the nonlinear case have been also been proposed in earlier work (Harmel-ing et al., 2003; Sprekeler et al., 2014).”
• Unsupervised Feature Extraction by Time-Contrastive Learning and Nonlinear ICA (2016, methodology): “We emphasize that our principle of nonstationarity is clearly distinct from the principle of linear autocorrelations previously used in the nonlinear case [12, 26].”
• Nonlinear ICA Using Auxiliary Variables and Generalized Contrastive Learning (2018, methodology): “In the case of Gaussian sources, PCL estimates the sources under the same conditions as the method by Sprekeler et al. (2014).”
• Nonlinear ICA of Temporally Dependent Stationary Sources (2017, methodology): “Even the general identifiability of the mixing models considered is not clear: There is no proof in the literature that the sources can be separated under the conditions considered — but see (Sprekeler et al., 2014) for some results in these directions, which are discussed in detail below.”
• Unsupervised Learning of Visual Representations using Videos (2015, methodology): “This method has been successfully applied to a number of applications such as estimating driving forces of a dynamical system (Wiskott, 2003), blind source separation (Sprekeler et al., 2014) and as feature extractor for reinforcement learning (Legenstein et al.”
• Incremental Slow Feature Analysis: Adaptive Low-Complexity Slow Feature Updating from High-Dimensional Input Streams (2012, background): “SFA has shown success in many problems and scenarios — extraction of driving forces of a dynamical system (Wiskott, 2003), nonlinear blind source separation (Sprekeler et al., 2010), as a preprocessor for reinforcement learning (Legenstein et al.”
• Contrastive Learning Inverts the Data Generating Process (2021, background): “In the spirit of existing literature on nonlinear ICA (Hyvärinen & Pajunen, 1999; Harmeling et al., 2003; Sprekeler et al., 2014; Hyvärinen & Morioka, 2016; 2017; Gutmann & Hyvärinen, 2012; Hyvärinen et al., 2019; Khemakhem et al., 2020a), we assume that the observations x ∈ X are generated by a”