Optimal unsupervised learning in a single-layer linear feedforward neural network

By Terence D. Sanger. 1989

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

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

In Representation & Self-Supervised Learning

Citations in the corpus

Local Aggregation for Unsupervised Learning of Visual Embeddings (2019, background): “Unsupervised learning methods span a very broad spectrum of approaches going back to the roots of artificial neural networks [53, 39, 2, 58, 27, 41, 28, 30, 25], and are too numerous to fully review here.”
View-tolerant face recognition and Hebbian learning imply mirror-symmetric neural tuning to head orientation (2016, positive): “Plausible modifications of the rule—involving added noise or inhibitory connections with similar neurons—yield additional eigenvectors [45, 38].”
Online Representation Learning with Multi-layer Hebbian Networks for Image Classification Tasks (2017, methodology): “Extensions of the Hebb’s rule such as the Oja’s rule [1] or the Generalized Hebbian rule, also called Sanger’s rule [2] proved particularly efﬁcient at tasks such as online dimensionality reduction.”
CAVIAR: A 45k Neuron, 5M Synapse, 12G Connects/s AER Hardware Sensory–Processing– Learning–Actuating System for High-Speed Visual Object Recognition and Tracking (2009, background): “This data shows how the retina output is filtered down by the convolution to a cloud mostly around the center of the matching input circle, and how the WTA events further reduce the data to just events centered on the programmed object.”
Kernel Methods and Machine Learning (2014, unknown): “[198, 229].”
Finding a roadmap to achieve large neuromorphic hardware systems (2013, methodology): “These approaches are built around fundamental continuous-time ANN algorithms on PCA algorithms (Oja, 1982; Sanger, 1989) as well as ICA built from non-linearities (Hyvärinen and Oja, 1997), each with grounding to talk about potential computation and applications coupled with approaches to build…”
Candid Covariance-Free Incremental Principal Component Analysis (2003, methodology): “We propose a new method, candid covariance-free IPCA (CCIPCA), based on the work of Oja and Karhunen [10] and Sanger [11].”
Iterative kernel principal component analysis for image modeling (2005, methodology): “There is, however, a theorem by Oja (for <Y Z ) and Sanger (for < % ) on the local convergence of the GHA [36, 43]6 that directly carries over to the KHA.”
Nonlinear principal component analysis using autoassociative neural networks (1991, methodology): “The concept of using a neural network with a bottleneck to concentrate information has been previously discussed in the context of " encoder/ decoder " problems (Ackley et al., 1985; Rumelhart et al., 1986; Sanger, 1989).”
Neural networks for pattern recognition (1995, methodology): “Such networks can be made to perform principal component analysis of the data (Oja, 1982, 1989; Linsker, 1988; Sanger, 1989), and furthermore it can be arranged that the weights converge to orthonormal vectors along the principal component directions.”
Discovering Discrete Distributed Representations (1990, unknown): “(1989), Sanger (1989).”
Incremental Online Learning in High Dimensions (2005, methodology): “Actually, this additional regression could be avoided by replacing p with u in Step 2d, similar to techniques used in principal component analysis ( Sanger, 1989 ).”