Two NIPS papers available

[Paul Munro]

The following two papers from our group can be downloaded.  The first 
paper is available only in postscript and the second is in both 
postscript and acrobat versions.  The URLs can be found below with the 
abstracts.

Paul Munro                                Internet: munro at sis.pitt.edu
SIS Bldg 735                              Voice:        412-624-9427
Department of Information Science         Fax (new #):  412-624-2788
University of Pittsburgh           
Pittsburgh PA   15260

Personal HTML page = http://www.pitt.edu/~pwm/


(To be in :Advances in Neural Information Processing Systems 12, edited by 
S. A. Solla, T. K. Leen, and K.-R. Mueller, MIT Press)


             Effects of spatial and temporal contiguity on the 
                    acquisition of spatial information
 
                  Thea Ghiselli-Crippa and Paul W. Munro

                    URL: www.pitt.edu/~pwm/nips99a.ps

  ABSTRACT
  Spatial information comes in two forms: direct spatial information (for
  example, retinal position) and indirect temporal contiguity information,
  since objects encountered sequentially are in general spatially close.  The
  acquisition of spatial information by a neural network is investigated
  here.  Given a spatial layout of several objects, networks are trained 
  on a prediction task.  Networks using temporal sequences with no direct 
  spatial information are found to develop internal representations that 
  have distances correlated with distances in the external layout.  The 
  influence of spatial information is analyzed by providing direct spatial
  information to the system during training that is either consistent 
  with the layout or inconsistent with it.  This approach allows 
  examination of the relative contributions os spatial and temporal 
  contiguity.




              LTD facilitates learning in a noisy environment

                      Paul Munro and Gerardina Hernandez

                      URL: www.pitt.edu/~pwm/nips99b.ps
                           www.pitt.edu/~pwm/nips99b.pdf

  ABSTRACT
  Long-term potentiation (LTP) has long been held as a biological
  substrate for associative learning. Recently, evidence has emerged
  that long-term depression (LTD) results when the presynaptic cell
  fires after the postsynaptic cell. The computational utility of LTD
  is explored here. Synaptic modification kernels for both LTP and
  LTD have been proposed by other laboratories based studies of one
  postsynaptic unit. Here, the interaction between time-dependent
  LTP and LTD is studied in small networks.