The new version of the fann library have been released. The new release includes automatic building and training of ANN's using the Cascade2 algorithm. The release includes many other changes as seen in the changelog for the release.
Fast Artificial Neural Network Library is a free open source neural network library, which implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks. Cross-platform execution in both fixed and floating point are supported. It includes a framework for easy handling of training data sets. It is easy to use, versatile, well documented, and fast. PHP, C++, .NET, Python, Delphi, Octave, Ruby, Pure Data and Mathematica bindings are available. A reference manual accompanies the library with examples and recommendations on how to use the library. A graphical user interface is also available for the library. ... read more
There is a competition at http://www.linuxonpower.com/ where you can win an Apple Power Mac G5 by porting and optimizing fann to PowerPC64.
Seems like a really good initiative, but you have to live in the U.S. in order to participate.
Fast Artificial Neural Network Library implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks. Cross-platform execution in both fixed and floating point are supported. It includes a framework for easy handling of training data sets. It is easy to use, versatile, well documented, and fast. PHP, Python and Mathematica bindings are available.... read more
Fast Artificial Neural Network Library (fann)
http://fann.sourceforge.net/
fann is implemented in ANSI C. The library implements multilayerfeedforward networks with support for both fully connected and sparseconnected networks. Fann offers support for execution in fixed pointarithmetic to allow for fast execution on systems with no floatingpoint processor. To overcome the problems of integer overflow, thelibrary calculates a position of the decimal point after training andguarantees that integer overflow can not occur with this decimalpoint.... read more
A mailing list for the fann library have benn created. The mailing list can be accessed by clicking the "Lists" link at the top of this page.
The primary goal of this mailing list is to let developers and users of the fann library discuss various topics concerning the library.
If you use the library or would like to contribute to the library I would highly recommend subscribing to this mailing list.... read more
Fast Artificial Neural Network Library (fann)
fann is implemented in ANSI C. The library implements multilayer
feedforward networks with support for both fully connected and sparse
connected networks. Fann offers support for execution in fixed point
arithmetic to allow for fast execution on systems with no floating
point processor. To overcome the problems of integer overflow, the
library calculates a position of the decimal point after training and
guarantees that integer overflow can not occur with this decimal
point.... read more
Fast Artificial Neural Network Library (fann)
fann is implemented in ANSI C. The library implements multilayer
feedforward networks with support for both fully connected and sparse
connected networks. Fann offers support for execution in fixed point
arithmetic to allow for fast execution on systems with no floating
point processor. To overcome the problems of integer overflow, the
library calculates a position of the decimal point after training and
guarantees that integer overflow can not occur with this decimal
point.... read more
There is a bug in version 1.0.3 (I will release a bugfix later today). For now, please include these two lines after constructing the ann:
fann_set_activation_function_hidden(ann, FANN_SIGMOID_STEPWISE);
fann_set_activation_function_output(ann, FANN_SIGMOID_STEPWISE);
Regards,
Steffen
Fast Artificial Neural Network Library (fann)
fann is implemented in ANSI C. The library implements multilayer
feedforward networks with support for both fully connected and sparse
connected networks. Fann offers support for execution in fixed point
arithmetic to allow for fast execution on systems with no floating
point processor. To overcome the problems of integer overflow, the
library calculates a position of the decimal point after training and
guarantees that integer overflow can not occur with this decimal
point.... read more
This thread describes how to use the fann library in the borland c++ compiler.
http://sourceforge.net/forum/forum.php?thread_id=975544&forum_id=323465
Fast Artificial Neural Network Library (fann)
fann is implemented in ANSI C. The library implements multilayer
feedforward networks with support for both fully connected and sparse
connected networks. Fann offers support for execution in fixed point
arithmetic to allow for fast execution on systems with no floating
point processor. To overcome the problems of integer overflow, the
library calculates a position of the decimal point after training and
guarantees that integer overflow can not occur with this decimal
point.... read more
It is nice to see that people are downloading the fann library, but I would like to receive feedback from people using the library.
Feedback can be posted on the sourceforge site, or can be sent to me by email at lukesky@diku.dk (please write something with fann in the subject, to avoid being deleted as SPAM).
Regards,
Steffen
Fast Artificial Neural Network Library is written in in ANSI C. The library implements multilayer feedforward ANNs, up to 150 times faster than other libraries.
The homepage for this document has just been published on http://fann.sourceforge.net/
The primary contents of this homepage is the html documentation for the library.
I have posted a few examples to the "Docs" section, and encourages others to do the same.
Regards,
Steffen
I will create a proper configure script in the near future. I have no experience in doing this, but I will try my best. If somebody has particular requests to the configure script, then now is the time speak.
Minor bugs fixed. Changelog:
1.0.1 2003-10-03
* Bugfix concerning steepness parameter
* Moved some files around
* Added the changelog
I found a minor bug when setting the steepness value, I will soon release a bugfix.
Furthermore I will be looking into setting up a configure script.
The library will be released sometime in the beginning of november 2003 together with a user's guide, and a scientific report documenting the development of the library.