The NN Website and the NN Directory
The website nn.cs.utexas.edu and the directory
cs:/u/nn and the various products and shared
tools of the NN Group. The idea is that the website is public, vs. the
directory is intended for internal use.
The website is maintained as part of the "AI-lab" website, using the
AI-lab-wide database for research area descriptions, researcher
homepages, projects, demos, publications, and software/data. The
database is maintained through the AI-lab admin interface, with an
"nn" login. The NN site pages have their own design and organization
that's different from other groups. There are also areas for
spotlights, NN meetings and these advice pages that are outside the
All of us should have our own home page up to date with a research
description and a picture. The database nicely cross-references
areas, people, projects, demos, publications, and software so that
those fields are automatically populated in your home page (and
similarly in all other pages). Whenever you have produced something
new, like a paper, demo, or software, please make an entry for it in
the database. The fields are usually clear enough, but there are also
instructions on what needs to be included and how---please make sure
you follow the instructions!
The cs:/u/nn directory has material that generally
requires running a program through a unix shell (like LaTeX, neural net
simulators, etc). The README files in each subdirectory should give you
a pretty good idea of what is available.
Anyone with a cs account can access material in this directory, but you
have to belong to the nn group to
change anything in it. This directory serves as an archive, and only
well-polished things should go in here. If you are planning on doing
some major reorganization, talk to Risto first.
However, feel free to contribute to the archive if you come across
something that should go in here. Also, if you find any errors or
out-of-date information in any of the readme files or elsewhere, feel
free to fix them.
If you add something, remember to update the appropriate README file
(usually in the same directory). Also, there might be pages in this
user's guide that should refer to it, so check that as well. If it is
something major (a new tool etc), also send email to nn to let people
know. If you plan to update something drastically, it might be a good
idea to move the old version to /u/nn/tmp/old rather than deleting it,
just in case. The directories and files should be nn writable: say
"chmod g+w file" and "chgrp nn file", or else say
"/u/nn/bin/updatepermis directory" to change an entire directory and its
subdirectories at once.
Some directories are marked "Write
access controlled by CVS" in the list below. These are the
directories whose contents are edited regularly; CVS helps to us keep
track of the various edits made by different people. Please
do not directly edit any of the
files in the CVS-controlled directories in /u/nn. Instead, check out
the corresponding CVS project (cvs checkout
project-name) into your own home directory, edit your
local copy of the files, and then commit the changes. (The
project-name is the CVS module name for the
project; it is usually the same as the directory name, e.g. "bin" for
/u/nn/bin.) These steps are all described in the README files in each
Here's a short summary of what's in /u/nn:
- "aliases" has the email alias for the nn group, as well as
aliases for various projects.
- "bibs" has various bibliography files. The file nn.bib is our
primary bibliographic database for cognitive connectionism and neural
networks. Write access controlled by CVS (project "bibs").
- "bin" has various useful shell scripts, e.g. for updating nn file
protections etc. Write access controlled by CVS (project "bin").
- "cvsroot" is the repository for our web pages and for various
software packages. Unless you really know what you are doing, do
not change any of these files by hand.
- "data" has various training data (such as NetTalk, character
recognition, hansards), or at least publically-available subsets of
large sets for evaluation purposes. The directory cs:/projects has
- "demos.other" contains various working demos that aren't quite
ready for the public pages but may be run locally and used in talks
etc. Most of them are in /u/risto/demos, but here are a few
others. The scripts that run them are intended for linux
laptops/desktops. See also LocalDemos
for instructions on how to run the demos in /u/nn/demos.
- "mail" stores email to the nn alias during previous years (since
8/21/94; current year is in "mailbox".
- "mailbox" is the archive of messages sent to the nn alias during
the current year.
- "models" has source code for various neural network models and
architectures, including handy copies of our own models (most of which
are also on the website), and other publicly available code.
- "old" has various directories that are probably no longer
- "projects" is an archive of various past projects that somebody
might want to pick up and continue some day.
- "public_html", alias "www-site" is a link to our World Wide Web
pages. Write access to subdirectory "pages" is controlled by CVS
- "simulators" has source code and executables for various old
general-purpose neural network simulators, including Miyata's
PlaNet, PDP-software of McClelland and Rumelhart, Stuttgart
Neural Network simulator... and also a number of domain simulators
such as Khepera mobile robot, boat simulator, etc. that can be used as
testbeds for reinforcement learning.
- "tex" has macros and examples for formatting papers, letters,
posters, and talks with latex. They may be useful if Overleaf is not
an option for some reason. Write access controlled by CVS (project
- "tmp" is a working directory. You can use if if you want to
e.g. share some files while you are working on a project with someone,
or if you just need more temporary file space. Create a subdirectory
for yourself (with your username as the directory name) and delete the
files once you no longer need them. This directory is periodically
cleaned up so don't store anything permanent there.
- "tools" has various research tools like cluster analysis,
principal component analysis, numerical recipes, statistical
Last modified: Sun Sep 25 15:10:33 PDT 2022