Winner-Relaxing Self-Organizing Maps

A new family of self-organizing maps, the winner-relaxing Kohonen algorithm, is introduced as a generalization of a variant given by Kohonen in 1991. The magnification behavior is calculated analytically. For the original variant, a magnification exponent of 4/7 is derived; the generalized version allows steering the magnification in the wide range from exponent 1/2 to 1 in the one-dimensional case, thus providing optimal mapping in the sense of information theory. The winner-relaxing algorithm requires minimal extra computations per learning step and is conveniently easy to implement.

• Publication year

  2002

• Venue

  Neural Computation

• Publication date

  2002-08-21

• Fields of study

  Biology, Mathematics, Physics, Computer Science

• Identifiers
  DOI 10.1162/0899766053491922arXiv cond-mat/0208414
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Magnification Control in Winner Relaxing Neural Gas

  2006cited by this paper
• Self-organized formation of topologically correct feature maps

  2004cited by this paper
• Winner-relaxing and winner-enhancing Kohonen maps: Maximal mutual information from enhancing the winner

  2003cited by this paper
• Kernel-Based Topographic Map Formation by Local Density Modeling

  2002cited by this paper
• Asymptotic Level Density of the Elastic Net Self-Organizing Feature Map

  2002cited by this paper
• Faithful Representations and Topographic Maps: From Distortion- to Information-Based Self-Organization

  2000cited by this paper
• Energy functions for self-organizing maps

  1999influential reference
• Do cortical maps adapt to optimize information density?

  1999cited by this paper
• Kernel-Based Equiprobabilistic Topographic Map Formation

  1998influential reference
• Magnification control in neural maps

  1998cited by this paper
• PHASE TRANSITIONS IN STOCHASTIC SELF-ORGANIZING MAPS

  1997influential reference
• Nonparametric density estimation and regression achieved with topographic maps maximizing the information-theoretic entropy of their outputs

  1997cited by this paper
• Asymptotic level density in topological feature maps

  1995cited by this paper
• Statistical-mechanical analysis of self-organization and pattern formation during the development of visual maps.

  1992cited by this paper
• Neural computation and self-organizing maps - an introduction

  1992influential reference
• Asymptotic level density for a class of vector quantization processes

  1991cited by this paper
• Statistical mechanics as the underlying theory of ‘elastic’ and ‘neural’ optimisations

  1990cited by this paper
• How to Generate Ordered Maps by Maximizing the Mutual Information between Input and Output Signals

  1989influential reference
• Adding a conscience to competitive learning

  1988cited by this paper
• Convergence properties of Kohonen's topology conserving maps: fluctuations, stability, and dimension selection

  1988cited by this paper
• An analogue approach to the travelling salesman problem using an elastic net method

  1987influential reference
• On the stationary state of Kohonen's self-organizing sensory mapping

  1986influential reference

• Self-organizing maps with information theoretic learning

  2015cites this paper
• The Application of Divergences in Prototype Based Vector Quantization

  2014cites this paper
• ADAPTIVE KERNEL SELF-ORGANIZING MAPS USING INFORMATION THEORETIC LEARNING

  2010cites this paper
• SINAI : Self-Organizing Neural Systems for Internet TIN 2006-07362

  2009cites this paper
• Self-organizing maps with dynamic learning for signal reconstruction

  2007cites this paper
• Self-organizing Maps with Refractory Period

  2007cites this paper
• Magnification Control in Self-Organizing Maps and Neural Gas

  2006influential citation
• Investigation of Topographical Stability of the Concave and Convex Self-Organizing Map Variant

  2006cites this paper