Learning competitive

The training process of a Kohonen network consists of a competitive learning procedure and can be summarized as follows ... 

Grossberg, S. (1987). Competitive learning From interactive activation to adaptive resonance. Cognitive Science, 11, 23-63. 

Training a Kohonen neural network with a molecular descriptor and a spectrum vector models the rather complex relationship between a molecule and an infrared spectrum. This relationship is stored in the Kohonen network by assigning the weights through a competitive learning technique from a suitable training set of... 

Ueda, N., Nakano, R. A new competitive learning approach based on an equidistortion principle for designing optimal vector quantizers. Neural Networks 7(8), 1211-1227 (1994) Young, F.W. Multidimensional scaling. Encyclopedia of Statistical Sciences 5, 649-659... 

Learning in neural networks happens by associative or competitive learning laws. In this context, learning means the following ... 

In the case of competitive learning, the distance between the input vector, X, is compared to the weight vector, w, by using an appropriate distance measure. Usually, the Euchdian distance is applied (c Eq. (5.87)). In detail, the following steps are followed ... 

Unsupervised Competitive Learning This learning algorithm is used if no information about the class membership of the training data vectors is available. The change of the weights at iteration t is updated by... 

Competitive learning operates on the basis of a next neighbor classifier... 

Supervised Competitive Learning If class assignments of vectors are feasible, the weights can be trained in a supervised fashion. If we denote the class membership of the neuron by Dj, then we... 

The learning algorithm for the Kohonen net operates in analogy to the competitive learning laws in Eqs. 8.19-8.21 ... 

What makes the difference between an associative and competitive learning law ... 

Color Quantization with Magnitude Sensitive Competitive Learning Algorithm... 

Keywords Color Vector quantization Competitive learning Neural networks SaUency Binarization... 

A subset of VQ algorithms comprises Competitive Learning (CL) methods, where a neural network model is used to find an approach of VQ calculation in an unsupervised way. Their advantage over other VQ algorithms is that CL is simple and easily parallelizable. Well known CL approaches are K-means [13] (including... 

Methods based in traditional competitive learning are focused on data density representation to be optimal from the point of view of reducing the Shannon s information entropy for the use of codewords in a transmission task. However it is not always desirable a codebook representation with direct proportion between its codeword density and the data density. For example, in the human vision system, the attention is attracted to visually salient stimuli, and therefore only scene locations sufficiently different from their surroundings are processed in detail. A simple framework to think about how Saliency may be computed in biological brains has been developed over the past three decades [10,19]. 

We propose the use of MSCL neural network, to train this 3-D dataset, taking into account the magnitude function, that can be defined to lead the training process of the palette to accomplish the desired task. This algorithm follows the general Competitive Learning steps ... 

Section 2 describes the Magnitude Sensitive Competitive Learning (MSCL) method. 

This paper has shown the capabilities of MSCL algorithm for Color Quantization. MSCL is a neural competitive learning algorithm, which includes a magnitude function as a modulation factor of the distance used for the unit competition. As other competitive methods, MSCL accomplishes a vector quantization of the data. However, unlike most of the competitive methods who are oriented to represent in more detail only those zones with higher data-density, the magnitude function in MSCL can address the competitive process to represent any region. 

---