@hackage hfann0.2

Haskell binding to the FANN library

  • Categories

    • License

      BSD-3-Clause

    • Maintainer

      Olivier Boudry <olivier.boudry@gmail.com>

    • Links

    • Versions

      • 0.4.2 Tue, 30 Jul 2013
      • 0.4.1 Thu, 11 Jul 2013
      • 0.4 Thu, 29 Dec 2011
      • 0.3 Thu, 15 Dec 2011
      • 0.2 Thu, 25 Jun 2009
      • 0.1 Mon, 9 Jun 2008

    HFANN: A Haskell interface to the FANN library (http://leenissen.dk/fann/)

    Copyright: 2008, Olivier Boudry Licence: BSD3

    Installation

    This package was developped and tested with GHC-6.8.2 on the Win32 platform and the FANN library version 2.0.

    FANN (C-library) installation:

    First of all you need to install the FANN library.

    Download version 2.0 of the library from:

    http://leenissen.dk/fann/download.php

    and build it according to the instruction found on:

    http://leenissen.dk/fann/html/files2/installation-txt.html

    On windows download the source version and build it using MinGW instead of using the provided Visual C++ 6.0 Project File. Building with Visual C++ would create DLLs and require the 'stdcall' calling convention which would not work.

    Build and installation should be pretty straightforward:

    ./configure
make
make install

    HFANN (Haskell library) installation:

    Edit the hfann.cabal file and adapt the 'include-dirs' and 'extra-lib-dirs' parameters to your installation. On non windows platform just blanking those two params should work as long as the library and includes are installed in the lib and include places.

    Build and install using Cabal:

    runghc Setup.lhs configure
runghc Setup.lhs build
runghc Setup.lhs haddock
runghc Setup.lhs install

    Note: if you have haddock < 2.0 skip the haddock step. src/HFANN/Data.hsc is not parsed properly using earlier versions of haddock.

    Using the library

    You will find an example for training and using and artificial neural network for the 'xor' function in the 'examples/xor' directory.

    Feedback

    Complaints, feature requests and bug reports to: olivier.boudry@gmail.com