Neuro-Fuzzy Systems Software
Go to homepage

NEFPROX - Neuro-Fuzzy Function Approximation

Beta Release of NFIDENT Available!
(this page is still under construction)

NFIDENT is software for approximating functions with fuzzy systems based on supervised learning. NFIDENT implements our NEFPROX approach and - to some extent - the ANFIS approach of J.-S. Roger Jang.

With NFIDENT you can learn fuzzy systems of the Mamdani-type (realized by NEFPROX) or the Sugeno-type (realized by ANFIS). You can learn both fuzzy rules and membership functions. There is no graphical user interface, but some output for GNUPLOT is created to plot membership functions or training results.

Currently NFIDENT 0.9 Beta is available. NFIDENT is written in plain C and is intended for Unix systems. However, it should be easy to make the required changes, such that it compiles under MS-DOS.

NFIDENT is free software under the GNU public licence and comes without any kind of warranty.
Download the source code of NFIDENT (compressed tar-file, 52 KB)

There is no documentation yet. Please download one of our papers (fns97.ps.gz) for information on the NEFPROX model. You can find it in our list of papers.

For information on ANFIS e.g. refer to
J.-S. Roger Jang: ANFIS: Adaptive-network-based fuzzy inference systems. IEEE Trans. Systems, Man & Cybernetics, vol. 23: 665-685, 1993.
An ANFIS implementation by J.-S. Roger Jang can be obtained from the CMU Artificial Intelligence Repository (The FTP address is ftp://ftp.cs.cmu.edu/user/ai/areas/fuzzy/systems/anfis)