new paper in neuroprose

[Wolfgang Maass]

FTP-host: archive.cis.ohio-state.edu
FTP-filename: /pub/neuroprose/maass.perspectives.ps.Z

The file maass.perspectives.ps.Z is now available for copying
from the Neuroprose repository. This is a 37-page long paper.
Hardcopies are not available.

   PERSPECTIVES OF CURRENT RESEARCH ABOUT THE COMPLEXITY OF            
                    LEARNING ON NEURAL NETS

                          by
                     Wolfgang Maass

           Institute for Theoretical Computer Science   
      Technische Universitaet Graz, A-8010 Graz, Austria
             email: maass at igi.tu-graz.ac.at   

                  
Abstract:
This is a survey paper, which discusses within the framework 
of computational learning theory the current state of knowledge 
and important open problems in three areas of research about the
complexity of learning on neural nets:

 -- Effient algorithms for neural nets that learn from mistakes

 -- Bounds for the number of examples needed to train neural nets

 -- PAC-learning on neural nets without a-priori assumptions about
    the learning problem

All relevant definitions are given in the paper, and no previous
knowledge about computational learning theory is assumed.


************************ How to obtain a copy ************************

Via Anonymous FTP:

unix> ftp archive.cis.ohio-state.edu
Name: anonymous
Password: (type your email address)
ftp> cd pub/neuroprose
ftp> binary
ftp> get maass.perspectives.ps.Z
ftp> quit
unix> uncompress maass.perspectives.ps.Z
unix> lpr  maass.perspectives.ps (or what you normally do to print PostScript)