Release Notes

Neural Network Based
System Identification Toolbox

Version 1.1

Department of Automation, Technical University of Denmark, June 20, 1997



This note contains important information on how the present toolbox is to be 
installed and the conditions under which it may be used. Please read it 
carefully before use.

The note should be sufficient for being able to make the essential portion of 
the toolbox functions work properly. However, to enhance performance a number of 
functions have been rewritten in C and in order to compile these, it is 
necessary to read the information about CMEX files found in the Matlab External 
Interface Guide as well as to have some familiarity with compiling C programs.


INSTALLING THE TOOLBOX
o       The toolbox is provided in two versions. One for MATLAB 4.2 and one
        for MATLAB 5. Both versions have been tested under UNIX on a 
        HP9000/735 and under MS-WINDOWS 3.1/MS-WINDOWS 95 on an IBM compatible
        Pentium PC.

o       Due to a bug in the "fprintf" function in MATLAB 4.2 for MS-Windows, 
        many of the functions will generate a very inconvenient output in the
        MATLAB command window. If this is to be avoided, replace "\r" in the
        line:

        fprintf('iteration # %i   PI = %4.3\r',iteration-1,PI);

        with "\n" in the m-files batbp, incbp, marq, marq2, marqlm, nnarmax1,
        nnarmax2, nniol, nnoe, nnssif, nnrarmx1, nnrarmx2, rpe. (it is always
        close to line 10 from the bottom). The problem has been solved in 
        MATLAB 5.  On UNIX systems there should not be any such problems 
        (it is also thought that the toolbox can be used on Macintosh computers).

o       The signal processing toolbox is required by the validation functions
        (nneval, nnvalid, ifvalid) to compute the correlation functions.
        Otherwise no additional toolboxes should be necessary.

o       All toolbox functions are implemented as plain m-files, but to enhance
        performance CMEX duplicates have been written for some of the most
        important functions. In the MATLAB 4.2 version the Makefile contains 
        the commands necessary for compiling the C-routines. If one is running
        MATLAB 4.2 under HP-UX, it should be possible simply to write
        >> !make
        in the MATLAB command window to invoke the compiler. If one is working
        on another platform it will be necessary to modify the Makefile
        accordingly. It is strongly recommended that the compilation be
        optimized with respect to execution speed as much as the compiler
        permits. Under MATLAB5 it might be necessary to copy the file 
        mexopts.sh to the working directory and modify it appropriately. To
        compile the MEX files under MATLAB 5 just type
        >> makemex
        in the MATLAB command window.
 

USING THE TOOLBOX
o       The checks for incorrect functions calls are not very thorough and
        consequently MATLAB will often respond with quite incomprehensible error
        messages when a function is passed the wrong arguments. When calling a
        CMEX-function, it may even cause MATLAB to crash. Hence, when using the
        CMEX-functions it may be a good idea to make extra copies of the m-files
        they are replacing (do not just rename the m-files since they are still
        read by the "help" command). One can then start by calling the
        m-functions first to make sure the call is correct.

o       The functions have been optimized with respect to speed rather than
        memory usage. For large network architectures and/or large data sets,
        memory problems may thus occur. If one is running the student edition of
        MATLAB one will almost certainly run into problems due to the rather
        small maximum size of the matrices allowed.
 

CONDITIONS/DISCLAIMER
By using the toolbox the user agrees to all of the following.

o       If one is going to publish any work where this toolbox has been used,
        please remember it was obtained free of charge and include a reference
        to the technical report: M. Norgaard:"Neural Network Based System
        Identification Toolbox," Tech. Report. 97-E-851, Department of
        Automation, Technical University of Denmark, 1997.

o       Magnus Norgaard and IAU do not offer any support for this product
        whatsoever. The toolbox is offered free of charge - take it or leave it!

o       The toolbox is copyrighted freeware by Magnus Norgaard/Department of
        Automation, DTU. It may be distributed freely unmodified. It is,
        however, not permitted to utilize any part of the software in commercial
        products without prior written consent of Magnus Norgaard, The
        Department of Automation, DTU.

o       THE TOOLBOX IS PROVIDED "AS-IS" WITHOUT WARRENTY OF ANY KIND, EITHER
        EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRENTIES
        OR CONDITIONS OF MECHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. IN
        NO EVENT SHALL MAGNUS NORGAARD AND/OR THE DEPARTMENT OF AUTOMATION BE
        LIABLE FOR ANY SPECIAL, INCIDENTAL, INDIRECT, OR CONSEQUENTIAL DAMAGES
        OF ANY KIND, OR DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA, OR
        PROFITS, WHETHER OR NOT MN/IAU HAVE BEEN ADVISED OF THE POSSIBILITY OF
        SUCH DAMAGES, AND/OR ON ANY THEORY OF LIABILITY ARISING OUT OF OR IN
        CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 

MATLAB is a trademark of The MathWorks, Inc.
Borland C++ is a trademark of Borland International.
MS-Windows is a trademark of Microsoft Coporation.


Trademarks of other companies and/or organizations mentioned in this 
documentation appear for identification purposes only and are the property of 
their respective companies and/or organizations.

ACKNOWLEDGEMENTS
The author wish to acknowledge Associate Professors Elbert Hendricks, Paul Haase 
Sorensen, and Ole Ravn from the Department of Automation and Niels Kjolstad 
Poulsen from the Department of Mathematical Modelling for "beta testing" the 
toolbox and for giving invaluable comments on preprints of the manual. 
Furthermore, as it also is clear from the reference list, the basis for most of 
the work comes from discussions with members of the neural network group, 
Electronics Institute, who have provided great insight into neural network 
theory. Some parts of the toolbox were developed during a stay at the 
Neuro-Engineering Group, NASA Ames Research Center, and the group members are 
gratefully acknowledged for their many comments on the toolbox.


June 20, 1997
Magnus Norgaard
Department of Automation, Building 326
Technical University of Denmark
2800 Lyngby
Denmark
e-mail: pmn@iau.dtu.dk