Contact: Haralambos Sarimveis
|Input format:||Plain text or Excel file|
|Output format:||Plain text|
|User-specified parameters:||The user needs to define one tuning parameter, namely the number of fuzzy sets that are utilized to partition each input dimension.|
|Reporting information:||The following information is reported: number of hidden nodes, hidden node centers, widths of Gaussian function, output weights, predictions on the training set, (residuals, sum of squared errors, root mean squared error, F- statistic, coefficient of determination in regression problems), (overall %accuracy and %accuracy for each individual class in classification problems).|
Fuzzy-means is a training method for Radial Basis Function (RBF) neural networks and is based on the fuzzy
partition of the input space, which is produced by defining a number of triangular fuzzy sets in the domain of
each input variable. The centers of these fuzzy sets form a multidimensional grid on the input space. A
rigorous selection algorithm chooses the most appropriate vertices on the grid, which are then used as the
hidden node centers in the resulting RBF network model. The so called “fuzzy-means” training method does
not need the number of centers to be fixed before the execution of the method. Due to the fact that it is a one-
pass algorithm, it is extremely fast, even in the case of a large database of input-output training data. The
method was originally developed for solving nonlinear regression problems. A variant of the method for solving
classification problems has also been developed.
The algorithm has been implemented in the Matlab programming environment. Translation into C++
programming language is under development. The input formats accepted are Excel files and plain text. The
output is plain text.
For further information, we refer to the original publications [SAR02], [SAR06].
Background (publication date, popularity/level of familiarity, rationale of approach, further comments)
Fuzzy means for regression, published 2002 [SAR02]. Fuzzy means for classification,
published 2006 [SAR06]. The idea behind the selection algorithm is to place the centers
in the multidimensional input space, so that the distance between any two center
locations is guaranteed to be greater than a lower limit, which is defined by the length
of the edges on the grid. At the same time, the algorithm assures that for any input
example in the training set there is at least one selected hidden node that is close
enough, according to an appropriately defined distance criterion.
Bias (instance-selection bias, feature-selection bias, combined instance-selection/feature-selection bias, independence assumptions?, ...)
Lazy learning/eager learning
Interpretability of models (black box model?, ...)
Black box model
Type of Descriptor:
External components: JOELib
Input format: Plain text or Excel file
Output format: Plain text
[SAR02] Sarimveis, H., Alexandridis, A., Tsekouras, G and Bafas, G, (2002). A fast and efficient algorithm for training radial basis function neural networks based on a fuzzy partition of the input space, Industrial & Engineering Chemistry Research, 41:751-759.
[SAR06] Sarimveis, H., Doganis, P., Alexandridis, A (2006). A classification technique based on radial basis function neural networks, Advances in Engineering Software, 37(4):218-221.
[MEL06] Melagraki, G. Afantitis Α., Sarimveis, H., Iglessi-Markopoulou, O., Alexandridis, A (2006). A novel RBF neural network training methodology to predict toxicity to Vibrio fischeri, Molecular Diversity, 10(2): 213-221.