[5-2a] FTP and Other Resources: Agents -- Planning
In addition to programs available free by anonymous ftp, we've
included some programs which are available by contacting the authors,
and some programs which charge a nominal fee.
Agent Modelling:
ANIMALS is a simulation system written by Toby Tyrrell,
<toby@castle.ed.ac.uk>, for his PhD thesis. The thesis examines the
problem of action selection when dealing with realistic, animal-like
situations: how to choose, at each moment in time, the most
appropriate out of a repertoire of possible actions. It includes a
description is given of a simulated environment which is an extensive
and detailed simulation of the problem of action selection for
animals. This simulated environment is used to investigate the
adequacy of several theories of action selection (from both ethology
and artificial intelligence) such as the drive model, Lorenz's
psycho-hydraulic model and Maes' spreading activation network, and
outlines deficiencies in each mechanism. Finally, it proposes a new
approach to action selection is developed which determines the most
appropriate action in a principled way, and which does not suffer from
the inherent shortcomings found in other methods. The thesis includes
a review and bibliography of existing work on action selection. The
thesis is available by anonymous ftp from
ftp.ed.ac.uk:/pub/lrtt/ [129.215.146.5]
as the files as.1.ps.Z, as.2.ps.Z, ..., and as.7.ps.Z.
The simulation software is also available from the same site, as the
file se.tar.Z. The simulation software was written in Suntools rather
than Xtools. It can be run only from SunView or OpenWindows. The
action selection problem modelled by the simulated environment
comprises 15 different `sub-problems' (getting food, reproducing, not
getting lost, being vigilant for predators, etc), many internal and
external stimuli, and 35 different low-level actions to select
between.
ViewGen (Viewpoint Generator) is a Prolog program that implements a
"Belief Ascription Algorithm" as described in Ballim and Wilks (see the
bibliography section on User Modelling). This can be seen as a form of
agent modelling tool, which allows for the generation of arbitrarily deep
nested belief spaces based on the system's own beliefs, and on beliefs
that are typically held by groups of agents. ViewGen is available by
anonymous ftp from
crl.nmsu.edu:/pub/non-lexical/ViewFinder [128.123.1.18] (user anonymous)
ftp.ims.uni-stuttgart.de:/pub/ballim [141.58.127.8] (user ftp)
as the file ViewGen.tar.Z. The theory of belief ascription upon which
it is based is described in detail in Ballim and Wilks, and a general
framework for attributing and maintaining nested propositional
attitudes is described in Afzal Ballim's dissertation which is
archived with the Viewgen program (in the files
ViewFinder-{A4/A5/US}.tar.Z,
the variable part indicating the format of the PostScript file).
The inheritance reasoner is in the file vf-hetis.tar.Z.
Implemented in Sicstus prolog, and hence easily convertible to
any Edinburgh-style prolog. Contact Afzal Ballim <afzal@divsun.unige.ch>
for more information.
Analogical Reasoning:
SME -- multivac.ils.nwu.edu:/pub/SME
Contact: Brian Falkenhainer <falkenhainer@parc.xerox.com>
Ken Forbus <forbus@ils.nwu.edu>
the Structure-Mapping Engine, as described in Falkenhainer,
Forbus, and Gentner's 1987 AIJ article.
Artificial Life:
Tierra is an artificial life system for studying the evolution of
digital organisms. Tierra consists of a virtual computer and its
operating system, whose architecture has been designed in such a way
that the executable machine codes are evolvable. This means that the
machine code can be mutated (by flipping bits at random) or recombined
(by swapping segments of code between algorithms), and the resulting
code remains functional enough of the time for natural (or presumably
artificial) selection to be able to improve the code over time.
Tierra runs in Unix and MS-DOS. Source code and documentation is
available by anonymous ftp at
tierra.slhs.udel.edu:/almond/ [128.175.41.34]
tierra.slhs.udel.edu:/beagle/
tierra.slhs.udel.edu:/doc/
tierra.slhs.udel.edu:/tierra/
or equivalently from
life.slhs.udel.edu:/almond/ [128.175.41.33]
life.slhs.udel.edu:/beagle/
life.slhs.udel.edu:/doc/
life.slhs.udel.edu:/tierra/
To be added to either the
tierra-announce (official announcements only) or
tierra-digest (moderated discussion plus announcements)
mailing lists, send mail to tierra-request@life.slhs.udel.edu.
Send bug reports to tierra-bug@life.slhs.udel.edu. Written by Tom Ray,
<ray@life.slhs.udel.edu>. [Tom's Current email is ray@hip.atr.co.jp;
mail to ray@udel.edu will be forwarded.] For those without access to
anonymous ftp, the Tierra software may be obtained on disk (DOS
executables) from Virtual Life, 25631 Jorgensen Rd., Newman, CA 95360.
Virtual Life will also be offering a PC version of Karl Sims'
evolutionary art system, called Darwinin Art.
Blackboard Architectures:
GBB (PD Version) -- ftp.cs.umass.edu:/gbb/
Case-based Reasoning:
CL-Protos -- cs.utexas.edu:/pub/porter/
Contact: Bruce W. Porter <porter@cs.utexas.edu>
Ray Bareiss <bareiss@ils.nwu.edu>
Erik Eilerts <eilerts@cs.utexas.edu>
Dan Dvorak
MICRO-xxx -- cs.umd.edu:/pub/schank/icbr/
Contact: waander@cs.umd.edu
The directory /pub/schank/icbr/ contains the complete
code for "Inside Case-Based Reasoning" by
Riesbeck and Schank, 1989. This includes code
for an instructional version of CHEF by Kristian Hammond.
Chess:
The SAN Kit chess programming C source toolkit provides common routines
for move notation I/O, move generation, move execution, etc. Only search
routines and an evaluation function need be added to obtain a working
chess program. It runs on Apple Macintosh (Think C 5.0),
Commodore Amiga (SAS C), MS-DOS, and Unix. It is available by
anonymous ftp from
raven.alaska.edu:/pub/coherent/sources32/ [137.229.10.39] in the
chess.lm.com:/pub/chess/Unix/
as the compressed tar file SAN.tar.Z or SAN.tar.gz.
Contact Steven J. Edwards <sje@world.std.com> for more information.
valkyries.andrew.cmu.edu:/pub/misc/chess/ [128.2.232.4]
This site has the SCP package, a restructured ANSI C port of
the 1987 Stanback Chess Program.
PGN (Portable Game Notation) is a specification for a standard move
notation system that has been adopted by many programs and toolkits.
It is available by anonymous ftp as
chess.lm.com:/pub/chess/PGN/Standard
Many chess-related materials are available from chess.lm.com:/pub/chess/
Complex Systems:
A list of resources for Complex Adaptive Systems is maintained by
Alex Mallet, including information about chaos theory, genetic
programming, artificial life, and neural networks. To get a copy by
email, send a message to
thesisnet-request@eniac.seas.upenn.edu
with
cplxsys
in the subject line. A hypertext version is available by WWW from
http://www.seas.upenn.edu/~ale/cplxsys.html
Send corrections to Alex Mallet <ale@eniac.seas.upenn.edu>.
Constraint Programming and Non-determinism:
SCREAMER:
Screamer is an extension of Common Lisp that adds support for
nondeterministic programming. Screamer consists of two levels. The
basic nondeterministic level adds support for backtracking and
undoable side effects. On top of this nondeterministic substrate,
Screamer provides a comprehensive constraint programming language in
which one can formulate and solve mixed systems of numeric and
symbolic constraints. Together, these two levels augment Common Lisp
with practically all of the functionality of both Prolog and
constraint logic programming languages such as CHiP and CLP(R).
Furthermore, Screamer is fully integrated with Common Lisp. Screamer
programs can coexist and interoperate with other extensions to Common
Lisp such as CLOS, CLIM and Iterate.
In several ways Screamer is more efficient than other implementations
of backtracking languages. First, Screamer code is transformed into
Common Lisp which can be compiled by the underlying Common Lisp
system. Many competing implementations of nondeterministic Lisp are
interpreters and thus are far less efficient than Screamer. Second,
the backtracking primitives require fairly low overhead in Screamer.
Finally, this overhead to support backtracking is only paid for those
portions of the program which use the backtracking primitives.
Deterministic portions of user programs pass through the Screamer to
Common Lisp transformation unchanged. Since in practise, only small
portions of typical programs utilize the backtracking primitives,
Screamer can produce more efficient code than compilers for languages
in which backtracking is more pervasive.
Screamer is fairly portable across most Common Lisp implementations.
It currently runs under Genera 8.1.1 and 8.3 on both Symbolics 36xx
and Ivory machines, under Lucid 4.0.2 and 4.1 on Sun SPARC machines,
under MCL 2.0 and 2.0p2 on Apple Macintosh machines, and under Poplog
Common Lisp on Sun SPARC machines. It should run under any
implementation of Common Lisp which is compliant with CLtL2 and with
minor revision could be made to run under implementations compliant
with CLtL1 or dpANS.
Screamer is available by anonymous FTP from
ftp.ai.mit.edu:/pub/screamer.tar.Z
Contact Jeffrey Mark Siskind <qobi@ai.mit.edu> for further information.
The Screamer Tool Repository, a collection of user-contributed
Screamer code, is available by anonymous ftp from
ftp.cis.upenn.edu:/pub/screamer-tools/
or by WWW from
http://www.cis.upenn.edu/~screamer-tools/home.html
Please direct all inquires about the repository to
screamer-repository@cis.upenn.edu.
Data Mining and Knowledge Discovery in Databases:
Explora is a data mining package written in Lisp for the Macintosh. It
includes a natural language hypertext-type interface for presentation
of dicoveries. It is available by anonymous FTP from
ftp.gmd.de:/GMD/explora/
as the files Explora.sit.hqx and README. For more information, see
http://orgwis.gmd.de:80/explora/
INSPECT is a PC-based data mining tool with visualization and neural
networks. It is available by anonymous FTP from
ftp.tuwien.ac.at:/Sources/NeuralNet/Inst-of-Chem/
See the readme.txt file for details.
Defeasible Reasoning:
An implementation of J. Paris and A. Vencovska's model of belief is
available by anonymous ftp from
ftp.cs.cmu.edu:/user/ai/areas/reasonng/defeasbl/belief/
Paris and Vencovska's paper (Artificial Intelligence, 64(2), December
1993) provides a mathematical model of an agent's belief in an event
by identifying it with his ability to imagine the event within the
context of his previous experience. This approach leads to beliefs
having properties different from those normally ascribed to it. The
implementation was written by Ian Pratt <ipratt@cs.man.ac.uk> and Jens
Doerpmund <dorpmunj@cs.man.ac.uk> and runs in Common Lisp.
Eliza and Similar Programs:
For a large collection of Eliza programs, see
ftp.cs.cmu.edu:/user/ai/areas/classics/
The software from Peter Norvig's book "Paradigms of AI Programming" is
available by anonymous ftp from unix.sri.com:/pub/norvig/ and on disk in
Macintosh or DOS format from the publisher, Morgan Kaufmann. The
software includes Common Lisp implementations of: Eliza and pattern
matchers, Emycin, Othello, Parsers, Scheme interpreters and compilers,
Unification and a prolog interpreter and compiler, Waltz
line-labelling, implementation of GPS, macsyma, and random number
generators. For more information, write to Morgan Kaufmann, Dept. P1,
2929 Campus Drive, Suite 260, San Mateo CA 94403, call 800-745-7323,
or fax 415-578-0672. (Mac ISBN 1-55860-227-5; DOS 3.5" ISBN
1-55860-228-3; or DOS 5.25" ISBN 1-55860-229-1).
The doctor.el is an implementation of Eliza for GNU-Emacs
emacs-lisp. Invoke it with "Meta-X doctor".
Source code for ELIZA in Prolog (implemented by Viren
Patel) is available by ftp from aisun1.ai.uga.edu.
muLISP-87 (a MSDOS Lisp sold by Soft Warehouse) includes
a Lisp implementation of Eliza.
Compute!'s Gazette, June 1984, includes source for a BASIC
implementation of Eliza. You can also find it in 101 more computer
games, edited by David Ahl, published by Creative Computing (alas,
they're defunct, and the book is out of print).
Herbert Schildt "Artificial Intelligence using C", McGraw-Hill, 1987,
ISBN 0-07-881255-0, pp315-338, includes a simple version of DOCTOR.
ucsd.edu:/pub/pc-ai contains implementations of Eliza for the IBM PC.
eecs.nwu.edu:/pub/eliza/ contains several Eliza implementations.
The original Parry (in MLISP for a PDP-10) is available in
labrea.stanford.edu:/pub/parry.tar.Z.
RACTER is *not* public domain. It costs $50 for MS-DOS and Macintosh
versions, the Inrac compiler is $200 (MS-DOS only), and the Inrac
manual alone is $25. Racter is available from John Owens, INRAC
Corp./Nickers International Ltd., 12 Schubert Street, Staten Island,
NY 10305, Tel: 718-448-6283, or Fax: 718-448-6298. Racter was
published in 1984, and written in compiled BASIC. To read some of
RACTER's work, see "The Policeman's Beard is Half Constructed",
Computer Prose and Poetry by Racter, Warner Books, 1984. ISBN
0-446-38051-2 (paperback). Written by William Chamberlain and Thomas
Etter. Some discussion of RACTER appears in A.K. Dewdney's book, "The
Armchair Universe". The Macintosh version runs only on SEs and Pluses
(it comes on a single-sided 400k copy-protected disk, with an old
version of the system). Racter is also sold by the following
mail-order software retailer: Mindware, 1803 Mission Street, Suite
414, Santa Cruz, CA 95060-5292, phone 800-447-0477 (408-427-9455),
fax 408-429-5302. Mindware sells a variety of similar programs for
MS-DOS and Windows, including Joseph Weintraub's PC Therapist.
Expert Systems:
Free ftpable expert system shells are listed in the Expert Systems
Shells FAQ, which is available by anonymous ftp from
ftp.cs.cmu.edu:/user/ai/pubs/faqs/ai/expert_1.faq
Frame Systems:
FrameWork -- ftp.cs.cmu.edu:/user/ai/areas/kr/frames/framework/
Theo -- Contact: Tom.Mitchell@cs.cmu.edu
FrameKit -- Contact: Eric.Nyberg@cs.cmu.edu
KR -- Contact: Brad.Myers@cs.cmu.edu
PARKA -- Contact: spector@cs.umd.edu
Frames for the CM
PARMENIDES (Frulekit) -- Contact: Peter.Shell@cs.cmu.edu
FROBS -- cs.utah.edu:/pub/frobs.tar.Z
Contact: Robert Kessler <kessler@cs.utah.edu>
PFC -- linc.cis.upenn.edu: ??
YAK -- Contact: Enrico Franconi <franconi@irst.it>
Fuzzy Logic:
FLIE -- ural.ethz.ch:/robo/flie/ [129.132.104.194]
Contact: vestli@ifr.ethz.ch
Fuzzy Logic Inference Engine, Institute of Robotics, ETH.
RICE (Routines for Implementing C Expert systems) is a fuzzy/MV logic
inference engine written in C. A C++ front-end with classes is provided.
Tested with Borland C/C++ 3.1, Microsoft C/C++ 7.00 and GCC 2.4.5;
examples are included. Documentation is available in WP 5.1 format and
PostScript. Available by anonymous ftp from ntia.its.bldrdoc.gov and
ftp.cs.cmu.edu. For more info contact Rene' Jager, <R.Jager@ET.TUDelft.NL>.
FuNeGen 1.0 is a fuzzy neural system capable of generating fuzzy
classification systems (as C-code) from sample data.
FuNeGen 1.0 and the papers/reports describing the application and the
theoretical background can be obtained by anonymous ftp from
obelix.microelectronic.e-technik.th-darmstadt.de:/pub/neurofuzzy/
Game Playing:
METAGAME is a game-playing workbench for developing and playing
metagame programs. It includes a generator for symmetric chess-like
games; definitions of chess, checkers, chinese chess, shogi, lose
chess, lose checkers, french checkers, and tic tac toe translated into
symmetric chess-like games; a legal move generator; and a variety of
player programs, from simple through sophisticated. The METAGAME
Workbench runs in Quintus or Sictus Prolog. Available by anonymous
ftp from
ftp.cl.cam.ac.uk:/users/bdp/metagame3a.tar.Z [128.232.0.56]
For more information, contact Barney Pell <bdp@cl.cam.ac.uk> of the
University of Cambridge Computer Laboratory.
Genetic Algorithms:
SCS (Simple Classifier System) is a C port of the system from
Appendix D of "Genetic Algorithms in Search, Optimization, and Machine
Learning" by David E. Goldberg. It was ported to C by Erik Mayer
<emayer@uoft02.utoledo.edu>. For more information, contact the author.
SCS-C is another port to C of Goldberg's Simple Classifier System.
It includes some extensions, and runs on Sun 10/30 and Atari ST. SCS-C
is available via anonymous ftp as scs-c-0.98j.tar.Z from
lumpi.informatik.uni-dortmund.de:/pub/LCS/src/ [129.217.36.140]. The
documentation alone is available as scs-c-doc.tar.Z in the directory
/pub/LCS/docs/. For more information, contact Joerg Heitkoetter
<joke@ls11.informatik.uni-dortmund.de>, c/o Systems Analysis Group,
LSXI, Department of Computer Science, University of Dortmund, D-44221
Dortmund, Germany.
GENITOR is available by anonymous ftp from the Colorado State
University Computer Science Department in
beethoven.cs.colostate.edu:/pub/GENITOR.tar [129.82.102.183]
For further information, contact starkwea@cs.colostate.edu or
mathiask@cs.colostate.edu. If these fail to work, contact
whitley@cs.colostate.edu.
Other packages are described in detail in Nici Schraudolph's survey
of free and commercial GA software (see the Genetic Algorithms
Repository in [5-1]). Some of the free ones from Nici's list are
summarized below. Many are available from the GA Repository.
GAucsd Genetic algorithms software
cs.ucsd.edu:/pub/GAucsd/GAucsd14.ps.Z [132.239.51.3]
Contact GAucsd-request@cs.ucsd.edu
To be put on a mailing list of GAucsd users, send
the message "add GAucsd" to listserv@cs.ucsd.edu.
GAbench Genetic algorithms benchmarks and test problems
cs.ucsd.edu:/pub/GAbench/
Thomas Kammeyer (tkammeye@cs.ucsd.edu)
EM Evolution Machine (EM)
ftp-bionik.fb10.tu-berlin.de:/pub/software/Evolution-Machine/
[130.149.192.50]
em_tc.exe (EM for Turbo C)
em_tcp.exe (EM for Turbo C++)
em_man.exe (the manual)
Joachim Born <born@max.fb10.tu-berlin.de>
Genie GA-based modeling/forecasting system
Lance Chambers <P_Stampoul@fennel.cc.uwa.oz.au>
GENOCOP GEnetic algorithm for Numerical Optimization for
COnstrained Problems. Optimizes function with any
number of linear constraints (equalities and inequalities)
Genetic-2 Optimization package for the linear transportation problem.
Genetic-2N Optimization package for the nonlinear transportation problem.
All three were developed by Zbigniew Michalewicz and are
described in detail in his book "Genetic Algorithms + Data
Structures = Evolution Programs", Springer Verlag,
August 1992.
unccsun.uncc.edu:/coe/evol/ [152.15.10.88]
(also known as ftp.uncc.edu)
Zbigniew Michalewicz <zbyszek@unccvax.uncc.edu>
WOLF Simulator for G/SPLINES algorithm (genetic spline models)
David Rogers <drogers@riacs.edu>
GAC, GAL GA written in C/Lisp. Similar to John Grefenstette's Genesis.
Bill Spears <spears@aic.nrl.navy.mil>
ESCaPaDE Experiments with evolutionary algorithsm.
Frank Hoffmeister <iwan@ls11.informatik.uni-dortmund.de>
(Send mail with subject line "help" or "get ESCaPaDE")
mGA1.0 Common Lisp implementation of a messy GA as described in
TCGA report 90004.
SGA-C C-language port and extension of the original Pascal
SGA code presented in Goldberg's book "Genetic
Algorithms in Search, Optimization & Machine
Learning", Addison Wesley, 1989. See TCGA report 91002.
SGA-Cube Goldberg's SGA code modified for nCUBE 2 hypercube
parallel computer.
All three are available by e-mail from
Robert Elliott Smith <rob@comec4.mh.ua.edu>.
BUGS Demonstrates genetic algorithms.
santafe.edu:/pub/misc/BUGS/
Joshua Smith <jrs@santafe.edu>
SGPC Simple Genetic Programming in C
sfi.santafe.edu:/pub/Users/tackett/
Walter Alden Tackett and Aviram Carmi (gpc@ipld01.hac.com)
GENEsYs lumpi.informatik.uni-dortmund.de:/pub/GA/src/ [129.217.36.140]
Use "ftp" as user name, e-mail address as password.
Thomas Baeck <baeck@ls11.informatik.uni-dortmund.de>
GAGA Jon Crowcroft <jon@cs.ucl.ac.uk>. cs.ucl.ac.uk:darpa/gaga.shar
Splicer Steve Bayer <bayer@galileo.jsc.nasa.gov>
PARAGENESIS GA-Repository/e-mail Michael van Lent <vanlent@cs.utk.edu>
GENESIS GA-Repository/e-mail John Grefenstette <gref@aic.nrl.navy.mil>
OOGA GA-Repository/e-mail John Grefenstette <gref@aic.nrl.navy.mil>
DGENESIS Erick Cantu <ecantu@babbage.rhon.itam.mx> or
<ecantu@itamvms1.bitnet>.
PGA Parallel Genetic Algorithms testbed
ftp.dai.ed.ac.uk:/pub/pga-2.4/pga-2.4.tar.Z (192.41.104.152)
Peter Ross, peter@aisb.ed.ac.uk
ANT PC Version of 'John Muir Trail' experiment.
ftp.std.com:/pub/pbrennan
Patrick M Brennan <pbrennan@world.std.com>
GPQUICK is a simple GP system implemented in C++. It features an
elegant object architecture with function (Function), program
(Chrome), GA (Pop) and problem (Problem) classes. The Problem class
is proposed as a portable representation for problems that would be
source compatible with a variety of other GP systems. GPQUICK uses a
steady state GA, tournament selection, one type of mutation, and
subtree crossover. It uses a fast, compact linear representation for
S-expressions. It includes documentation from the associated magazine
article (Byte, "Some Assembly Required", February 1994). GPQUICK is
available by anonymous ftp from
ftp.cc.utexas.edu:/pub/genetic-programming/code/
as the files gpquick.tar (unix version, tested with CC and g++) and
gpquick.zip (PC/ANSI C version, tested with Borland 3.1).
For more information, write to Andrew Singleton <p00396@psilink.com>.
GENlib is a library of functions for genetic algorithms together with
two applications of the library to train neural networks. The library
is available by anonymous ftp from
ftp.neuro.informatik.uni-kassel.de:/pub/NeuralNets/GA-and-NN/
for academic research and educational purposes only. Commercial uses
require written permission from the author. For more information,
write to Jochen Ruhland <jochenr@neuro.informatik.uni-kassel.de>.
ICOT:
Japan's Institute for New Generation Computer Technology (ICOT) has
made their software available to the public free of charge. The
collection includes a variety of prolog-based programs in symbol
processing, knowledge representation, reasoning and problem solving,
natural language processing. All programs are available by anonymous
ftp from ftp.icot.or.jp. Note that most of the programs are written
for the PSI machines, and very few have been ported to Unix-based
emulators. For further information, send email to ifs@icot.or.jp, or
write to ICOT Free Software Desk, Institute for New Generation
Computer Technology, 21st Floor, Mita Kokusai Bldg., 4-28, Mita
1-chome, Minato-ku, Tokyo 108, Japan, fax +81-3-4456-1618.
Knowledge Representation:
KNOWBEL -- ai.toronto.edu:/pub/kr/ as the files knowbel.tar.Z and
manual.txt.tar.Z
Contact: Bryan M. Kramer, <kramer@ai.toronto.edu>
Telos temporal/sorted logic system.
SB-ONE -- Contact: kobsa@inf-wiss.uni-konstanz.de
KL-ONE family. Currently undergoing revision and will be
renamed KN-PART+.
KRIS -- Contact: baader@dfki.uni-kl.de
KL-ONE family (Symbolics only)
BACK -- Contact: back@cs.tu-berlin.de
ftp.cs.tu-berlin.de:/pub/doc/reports/tu-berlin.de/kit/Back52
Files are BACK_V52.intro and Back52.tar.Z
Tar file includes Tutorial/Manual in postscript format
and installation instructions.
KL-ONE family
CLASSIC -- Contact: dlm@research.att.com
KL-ONE family
MOTEL -- Contact: hustadt@mpi-sb.mpg.de
mpi-sb.mpg.de:/pub/tools/motel.tar.Z [139.19.1.1]
Modal KL-ONE (contains KRIS as a kernel).
Implemented in Prolog.
FOL GETFOL -- Contact: fausto@irst.it
Weyrauch's FOL system
COLAB/RELFUN -- Contact: boley@informatik.uni-kl.de
Logic Programming
COLAB/FORWARD -- Contact: hinkelma@dfki.uni-kl.de
Logic Programming
COLAB/CONTAX -- Contact: meyer@dfki.uni-kl.de
Constraint System for Weighted Constraints over
Hierarchically Structured Finite Domains.
COLAB/TAXON -- Contact: hanschke@dfki.uni-kl.de
Terminological Knowl. Rep. w/Concrete Domains
SNePS (Semantic Network Processing System) is the implementation of a
fully intensional theory of propositional knowledge representation and
reasoning. SNePS includes a module for creating and accessing
propositional semantic networks, path-based inference, node-based
inference based on SWM (a relevance logic with quantification) that
uses natural deduction and can deal with recursive rules, forward,
backward and bi-directional inference, nonstandard logical connectives
and quantifiers, an assumption based TMS for belief revision (SNeBR), a
morphological analyzer and a generalized ATN (GATN) parser for parsing
and generating natural language, SNePSLOG, a predicate-logic-style
interface to SNePS, XGinseng, an X-based graphics interface for
displaying, creating and editing SNePS networks, SNACTor, a
preliminary version of the SNePS Acting component, and SNIP 2.2, a new
implementation of the SNePS Inference Package that uses rule shadowing
and knowledge migration to speed up inference. SNeRE (the SNePS
Rational Engine), which is part of Deepak Kumar's dissertation about
the integration of inference and acting, will replace the current
implementation of SNACTor. SNePS is written in Common Lisp, and has
been tested in Allegro CL 4.1, Lucid CL 4.0, TI Common Lisp, CLISP
May-93, and CMU CL 17b. It should also run in Symbolics CL, AKCL 1.600
and higher, VAX Common Lisp, and MCL. The XGinseng interface is built
on top of Garnet. SNePS 2.1 is free according to the GNU General
Public License version 2. The SNePS distribution is available by
anonymous ftp from
ftp.cs.buffalo.edu:/pub/sneps/ [128.205.32.9]
as the file rel-x-yyy.tar.Z, where 'x-yyy' is the version. The other
files in the directory are included in the distribution; they are
duplicated to let you get them without unpacking the full distribution
if you just want the bibliography or manual. If you use SNePS, please
send a short message to shapiro@cs.buffalo.edu and
snwiz@cs.buffalo.edu. Please also let them know whether you'd like to
be added to the SNUG (SNePS Users Group) mailing list.
URANUS is a logic-based knowledge representation language. Uranus is
an extension of Prolog written in Common Lisp and using the syntax of
Lisp. Uranus extends Prolog with a multiple world mechanism for
knowledge representation and term descriptions to provide
functional programming within the framework of logic programming.
It is available free by anonymous ftp from
etlport.etl.go.jp:/pub/uranus/ftp/ [192.31.197.99]
for research purposes only. For more information contact the author,
Hideyuki Nakashima <nakashim@etl.go.jp>.
Machine Learning:
COBWEB/3 -- Contact: cobweb@ptolemy.arc.nasa.gov
IND -- Contact: NASA COSMIC, <service@cossack.cosmic.uga.edu>
Tel: 706-542-3265 (ask for customer support)
Fax: 706-542-4807
IND is a C program for the creation and manipulation of
decision trees from data, integrating the CART,
ID3/C4.5, Buntine's smoothing and option trees, Wallace
and Patrick's MML method, and Oliver and Wallace's MML
decision graphs which extend the tree representation to
graphs. Written by Wray Buntine, <wray@kronos.arc.nasa.gov>.
AUTOCLASS -- Contact: taylor@ptolemy.arc.nasa.gov
AutoClass is an unsupervised Bayesian classification system for
independent data.
FOIL -- ftp.cs.su.oz.au:/pub/ [129.78.8.208]
as the files foil4.sh, foil5.sh, and foil6.sh.
Each shell archive contains source, a brief manual,
and several sample datasets. FOIL2 should be available
from sumex-aim.stanford.edu:/pub/FOIL.sh. FOIL 6.0 now uses
ANSI C.
Contact: J. Ross Quinlan <quinlan@cs.su.oz.au>
Mike Cameron-Jones <mcj@cs.su.oz.au>
RWM -- Contact: H. Altay Guvenir <guvenir@trbilun.bitnet>
RWM is a program for learning problem solving strategies,
written in Common Lisp (tested on Suns and NeXT).
MOBAL is a system for developing operational models of application
domains in a first order logic representation. It integrates a manual
knowledge acquisition and inspection environment, an inference engine,
machine learning methods for automated knowledge acquisition, and a
knowledge revision tool. By using MOBAL's knowledge acquisition
environment, you can incrementally develop a model of your domain in
terms of logical facts and rules. You can inspect the knowledge you
have entered in text or graphics windows, augment the knowledge, or
change it at any time. The built-in inference engine can immediately
execute the rules you have entered to show you the consequences of
your inputs, or answer queries about the current knowledge. MOBAL also
builds a dynamic sort taxonomy from your inputs. If you wish, you can
use several machine learning methods to automatically discover
additional rules based on the facts that you have entered, or to form
new concepts. If there are contradictions in the knowledge base due to
incorrect rules or facts, there is a knowledge revision tool to help
you locate the problem and fix it. MOBAL (release 3.0b) is available
free for non-commercial academic use by anonymous ftp from
ftp.gmd.de:/gmd/mlt/Mobal/
The system runs on Sun SparcStations, SunOS 4.1, and includes a
graphical interface implemented using Tcl/TK.
PEBLS (Parallel Exemplar-Based Learning System) is a nearest-neighbor
learning system designed for applications where the instances have
symbolic feature values. PEBLS has been applied to the prediction of
protein secondary structure and to the identification of DNA promoter
sequences. PEBLS 3.0 is written in ANSI C and is available by
anonymous ftp from blaze.cs.jhu.edu:/pub/pebls/pebls.tar.Z
[128.220.13.50] for research purposes only. For more information,
contact Steven Salzberg <salzberg@cs.jhu.edu>.
OC1 (Oblique Classifier 1) is a multivariate decision tree induction
system designed for applications where the instances have numeric
feature values. OC1 builds decision trees that contain linear
combinations of one or more attributes at each internal node; these
trees then partition the space of examples with both oblique and
axis-parallel hyperplanes. OC1 has been used for classification of
data from several real world domains, such as astronomy and cancer
diagnosis. A technical decription of the algorithm can be found in
the AAAI-93 paper by Sreerama K. Murthy, Simon Kasif, Steven Salzberg
and Richard Beigel. A postscript version of this paper is included in
the distribution. OC1 is a written entirely in ANSI C. OC1 is
available by anonymous ftp from
blaze.cs.jhu.edu:/pub/oc1/ [128.220.13.50]
This distribution is provided for non-commercial purposes only. For
more information, contact Sreerama K. Murthy <murthy@cs.jhu.edu>
(primary contact), Steven Salzberg <salzberg@cs.jhu.edu>, or Simon
Kasif <kasif@cs.jhu.edu>, Department of Computer Science, The Johns
Hopkins University, Baltimore, MD 21218.
Set-Enumeration (SE) Trees for Induction/Classification. Significant
research in Machine Learning, and in Statistics, has been devoted to
the induction and use of decision trees as classifiers. An induction
framework which generalizes decision trees using a Set-Enumeration
(SE) tree is outlined in
Rymon, R. (1993), An SE-tree-based Characterization of the Induction
Problem. In Proc. of the Tenth International Conference on Machine
Learning, Amherst MA, pp. 268-275.
In this framework, called SE-Learn, rather than splitting according to
a single attribute, one recursively branches on all (or most) relevant
attributes. An induced SE-tree can be shown to economically embed many
decision trees, thereby supporting a more expressive hypothesis
representation. Also, by branching on many attributes, SE-Learn
removes much of the algorithm-dependent search bias. Implementations
of SE-Learn can benefit from many techniques developed for decision
trees (e.g., attribute-selection and pruning measures). In particular,
SE-Learn can be tailored to start off with one's favorite decision
tree, and then improve upon it by further exploring the SE-tree. This
hill-climbing algorithm allows trading time/space for added accuracy.
Current studies (yet unpublished) show that SE-trees are particularly
advantageous in domains where (relatively) few examples are available
for training, and in noisy domains. Finally, SE-trees can provide a
unified framework for combining induced knowledge with knowledge
available from other sources (Rymon, 1994).
Rymon, R. (1994), On Kernel Rules and Prime Implicants. To appear in
Proc. of the Twelfth National Conference on Artificial Intelligence,
Seattle WA.
A Lisp implementation of SE-Learn is available from Ron Rymon
<Rymon@ISP.Pitt.edu>. A commercial version in C is currently under
development.
MLC++ is a Machine Learning library of C++ classes being developed at
Stanford. More information about the library can be obtained at URL
">robotics.stanford.edu:/users/ronnyk/mlc.html">http://robotics.stanford.edu:/users/ronnyk/mlc.html
The utilities are available by anonymous ftp from
starry.stanford.edu:/pub/ronnyk/mlc/util/
They are currently provided only as object code for Sun, but source code
will be distributed to sites that wish to port the code to other compilers.
For more information write to Ronny Kohavi <ronnyk@CS.Stanford.EDU>.
Mathematics:
SymbMath is a "symbolic calculator that can solve symbolic math
problems" written by Weiguang Huang <w.huang@unsw.edu.au>. It runs on
IBM PCs (8086) under MS-DOS. Shareware versions are available by
anonymous ftp from
wsmr-simtel20.army.mil:/calculator/sm22a.zip
rana.cc.deakin.oz.au:/huang/sm22a.zip
from the URL
http://acsusun.acsu.unsw.edu.au/~s9300078/symbmath.html
or by e-mail from listserv@vm1.nodak.edu (listserv@ndsuvm1.bitnet). To
subscribe to the symbmath@explode.unsw.edu.au mailing list, send email
to majordomo@explode.unsw.edu.au with
subscribe symbmath
in the message body.
Medical Reasoning:
TMYCIN -- sumex-aix.stanford.edu:/tmycin
Natural Language Processing:
YACC -- ftp.cs.cmu.edu:/user/ai/lang/lisp/code/parsing/lalr/
Contact: Mark Johnson <mj@cs.brown.edu>
Lisp YACC/Parser.
BABBLER -- Contact: rsf1@ra.msstate.edu
Markov chains/NLP
PENMAN -- Contact: hovy@isi.edu
Natural Language Generation.
PC-KIMMO -- msdos.archive.umich.edu:/msdos/linguistics/pckim105.zip
An implementation of KIMMO morphological analyzer
for the IBM PC.
FUF -- Contact: elhadad@bengus.bgu.ac.il
ftp: black.bgu.ac.il:/pub/fuf/fuf5.2.tar.Z
cs.columbia.edu:/pub/fuf/fuf5.2.tar.Z
Natural language generation system based on
Functional Unification Grammars.
Includes unifier, large grammar of English (surge)
user manual and many examples. Written in Common Lisp.
[A WAM-based C compiler for FUF is in the works.]
InterBASE -- Contact: Sergei Kuchin <kuchin@darmstadt.gmd.de>
ftp: files interbas.exe, interba1.exe, interbas.txt on
sics.se:/pub/packet-incoming
ftp.uu.net:/tmp
clr.nmsu.edu:/incoming
debra.dgbt.doc.ca:/pub/incoming
Natural language database front end
RegEx -- csd4.csd.uwm.edu:/pub/compilers/regex/
Translates regular expressions to DFAs. Written in C.
Mark Hopkins <markh@csd4.csd.uwm.edu>
Tom -- csd4.csd.uwm.edu:/pub/compilers/tomita/
C implementation of the Tomita parsing algorithm
Mark Hopkins <markh@csd4.csd.uwm.edu>
Common Lisp versions of the miniature natural language understanding
programs from "Inside Computer Understanding" by Schank and Riesbeck,
1981, are available by anonymous ftp from
cs.umd.edu:/pub/schank/icu/
This includes the SAM and ELI miniatures. It will
eventually include copies of the miniature versions of PAM, POLITICS,
and Tale-Spin. The FOR macro is also available in this directory, as
are a set of functions for manipulating and matching lisp
representations of Conceptual Dependency formulas. Contact Bill
Andersen <waander@cs.umd.edu> for more information.
The Link Parser is a highly efficient English parser written by Danny
Sleator and Davy Temperley. It uses a novel grammatical formalism known
as Link Grammar to represent a robust and diverse collection of
English-language phenomena. The system is available by anonymous ftp from
ftp://ftp.cs.cmu.edu/user/sleator/link-grammar/
Read the README file for more information. To see an online demo of
the parser, visit
http://bobo.link.cs.cmu.edu/cgi-bin/grammar/build-intro-page.cgi
Further information can be found on Danny Sleator's web page,
http://www.cs.cmu.edu/~sleator
The Xerox part-of-speech tagger is available by anonymous ftp from
parcftp.xerox.com:/pub/tagger/tagger-1-0.tar.Z. It is implemented in
Common Lisp and has been tested in Allegro CL 4.1, CMU CL 16e, and
Macintosh CL 2.0p2. For more information, contact the authors, Doug
Cutting <cutting@parc.xerox.com>, and Jan Pedersen
<pedersen@parc.xerox.com>.
Eric Brill's trainable rule-based part of speech tagger (version 1.0.2)
is available by anonymous ftp from
ftp.cs.jhu.edu:/pub/BRILL/Programs/
This tagger is based on transformation-based error-driven learning, a
technique that has been effective in a number of natural language
applications, including part of speech and word sense tagging,
prepositional phrase attachment, and syntactic parsing. For more
information, you can obtain relevant papers in
ftp.cs.jhu.edu:/pub/BRILL/Papers/
If you do download the tagger and wish to be on the mailing list for
future releases, bug reports, etc, please send mail to Eric Brill
<brill@cs.jhu.edu> or <brill@goldilocks.lcs.mit.edu>.
The Prolog and DCG programs from Pereira and Shieber's book, "Prolog
and Natural Language Analysis", are available by anonymous ftp from
das.harvard.edu:/pub/shieber/pnla/. See the file README for the
conditions under which the material is distributed. If you retrieve
the files, please send an email message to the authors letting them
know how you plan to use them. For further information, write to
Fernando Pereira <pereira@research.att.com> or Stuart Shieber
<shieber@das.harvard.edu>.
LHIP is a left-head-corner island parser compiler. The system compiles
grammar rules to Prolog code in much the same way as the Prolog DCG
system does. The rules themselves are an extended version of the DCG
rules, allowing optional constituents, negation, disjunction, the
specification of adjacency, and the ability to mark multiple heads in
a rule body. It requires an Edinburgh style Prolog and is known to
work in Sicstus 0.6. LHIP may be retrieved by anonymous ftp from
issun14.unige.ch:/pub/lhip_v1.1.tar.Z [129.194.177.14]
A more efficient version withou negation is also available:
issun14.unigh.ch:/pub/plhip_v1.0.tar.Z [129.194.177.14]
Both are also available from:
ftp.cs.cmu.edu:/user/ai/areas/nlp/parsing/lhip/lhip_v10.tar.gz
Please send a message to the author, Afzal Ballim <afzal@divsun.unige.ch>,
to let him know that you're using the package.
PAPPI is a Prolog-based natural language parser for theories in the
Principles-and-Parameters framework. The PAPPI system includes an
X Windows user interface and a sample implementation of classic GB
theory. PAPPI is available by anonymous FTP from
external.nj.nec.com:/pub/sandiway/Pappi-2.0f.tar.gz
For more information, please contact Dr. Sandiway Fong
<sandiway@research.nj.nec.com>.
Hdrug is an environment to develop logic grammars, parsers, and
generators for natural languages. The package comes with a number of
example grammars, including a Categorial Grammar, a Tree Adjoining
Grammar, a Unification Grammar in the spirit of Head-driven Phrase
Structure Grammar, an Extraposition Grammar, a Definite Clause
Grammar, and a port of the HPSG grammar from Bob Carpenter's ALE
system. Each of the grammars comes with a set of parsers, such as
Earley-like chart parsers, left-corner parsers and head-driven
parsers. Some grammars come with variants of the head-driven
generator. The package allows easy comparison of different
parsers/generators, extensive possibilities of compiling feature
equations into Prolog terms, graphical (Tk), LaTeX and ordinary Prolog
output of trees, feature structures and Prolog terms, and plotted
graphs and tables of statistical information. Hdrug runs in Sicstus
Prolog and requires ProTcl and Tcl/Tk. It is available by anonymous
FTP from
tyr.let.rug.nl:/pub/prolog-app/Hdrug/
or by WWW from
http://tyr.let.rug.nl/~vannoord/prolog-app/Hdrug/
For more information, write to Gertjan van Noord <vannoord@let.rug.nl>.
Neural Networks:
A draft review of roughly 40 neurosimulators is available by anonymous
ftp from
ftp.mrc-apu.cam.ac.uk:/pub/nn/
as the file neurosim1.ps.Z (text version in neurosim1.txt and
WordPerfect 5.1 version in neurosim1.w51.Z). The review will appear in
the "Handbook of Brain Research and Neural Networks" (MIT Press, 1995).
Please send comments to Dr. Jacob M.J. Murre <jaap.murre@mrc-apu.cam.ac.uk>.
Aspirin/MIGRAINES is a neural network simulator available free from the
MITRE Corporation. It contains a neural network simulation code generator
which generates high performance C code implementations for
backpropagation networks. It runs on the following platforms: Apollo,
Convex, Cray, DecStation, HP, IBM RS/6000, Intel 486/386 (Unix System V),
NeXT, News, Silicon Graphics Iris, Sun3, Sun4, Mercury i860 (40MHz)
Coprocessors, Meiko Computing Surface w/i860 (40MHz) Nodes, Skystation
i860 (40MHz) Coprocessors, and iWarp Cells. The software is available by
anonymous ftp from the CMU simulator collection on pt.cs.cmu.edu
(128.2.254.155) in the directory /afs/cs/project/connect/code (you must
cd to this directory in one atomic operation) and UCLA's cognitive
science collection on ftp.cognet.ucla.edu [128.97.8.19] in the
directory alexis as the file am6.tar.Z, am6.readme, am6.notes. They
include many examples in the release, include an implementation of NETtalk.
For more information, contact Russell Leighton <taylor@world.std.com>
or <leighton@mitre.org>. [As of 7/7/93, the mitre email address bounced.]
MUME (Multi-Module Neural Computing Environment) is a simulation
environment for multi-modules neural computing. It provides an object
oriented facility for the simulation and training of multiple nets
with various architectures and learning algorithms. The object
oriented structure makes simple the addition of new network classes
and new learning algorithms. MUME includes a library of network
architectures including feedforward, simple recurrent, and
continuously running recurrent neural networks. Each architecture is
supported by a variety of learning algorithms, including backprop,
weight perturbation, node perturbation, and simulated annealing. MUME
can be used for large scale neural network simulations as it provides
support for learning in multi-net environments. It also provide pre-
and post-processing facilities. MUME can be used to include
non-neural computing modules (decision trees, etc.) in applications. _
MUME is being developed at the Machine Intelligence Group at Sydney
University Electrical Engineering. The software is written in 'C' and
is being used on Sun and DEC workstations. Efforts are underway to
port it to the Fujitsu VP2200 vector processor using the VCC
vectorising C compiler, HP 9000/700, SGI workstations, DEC
Alphas, and PC DOS (with DJGCC). MUME is available to research
institutions on a media/doc/postage cost arrangement after
signing a license agreement. The license agreement is available by
anonymous ftp from mickey.sedal.su.oz.au:/pub/license.ps [129.78.24.170].
An overview of mume is available from the same machine as
/pub/mume-overview.ps.Z. It is also available free for MSDOS by
anonymous ftp from
brutus.ee.su.oz.au:/pub/MUME-0.5-DOS.zip
For further information, write to Marwan Jabri, SEDAL, Sydney
University Electrical Engineering, NSW 2006 Australia,
call +61-2-692-2240, fax +61-2-660-1228, or send email to
Marwan Jabri <marwan@sedal.su.oz.au>. To be added to the mailing
list, send email to mume-request@sedal.su.oz.au.
Adaptive Logic Network (ALN)
The atree adapative logic network simulation package is available by
anonymous ftp from
ftp.cs.ualberta.ca:pub/atree/ [129.128.4.241]
as the file atree2.tar.Z (Unix). The MS-Windows 3.x version for the
IBM PC is available as either atre27.exe (includes C/C++ sources) or
a27exe.exe (just the executables). The PC version has a lot more
documentation than the Unix version. The Unix version has been ported
to the Macintosh, Amiga, and other machines. Documentation is in
atree2.ps.Z. Also in this directory is a rather impressive OCR demo
using atree. To be added to the mailing list, send email to
alnl-request@cs.ualberta.ca. For more information, contact William W.
Armstrong, <arms@cs.ualberta.ca>.
BPS
Neural network simulator. Other files of interest. Executables are
free; source code for a small fee.
gmuvax2.gmu.edu:nn [no longer there?]
NeuralShell
Availible by anonymous ftp from
quanta.eng.ohio-state.edu:/pub/NeuralShell/ [128.146.35.1]
as the file NeuralShell.tar. [No longer available, due to an
alleged trademark infringement.]
CONDELA
A neural network definition language.
tut.cis.ohio-state.edu:/pub/condela
ROCHESTER CONNECTIONIST SIMULATOR
Available from ftp.cs.rochester.edu:/pub/packages/simulator [192.5.53.209].
Includes a backprop package and an X11/SunView interface.
UCLA-SFINX
retina.cs.ucla.edu:/pub/sfinx_v2.0.tar.Z [131.179.16.6]
Use username sfinxftp, password joshua. Contact sfinx@retina.cs.ucla.edu
for more information.
XERION
A neural network simulator from Drew van Camp at the University
of Toronto. It provides a library of routines for building networks
and graphically displaying them. Written in C and uses the X window
system for graphics. Example simulators include Back Propagation,
Recurrent Back Propagation, Boltzmann Machine, Mean Field Theory, Free
Energy Manipulation, Kohonnen Net, and Hard and Soft Competitive
Learning. Xerion runs on SGI Personal Iris, SGI 4d, Sun3 (SunOS), Sun4
(SunOS). Available by anonymous ftp from
ai.toronto.edu:/pub/xerion/
See the file /pub/xerion.README for more information. Also included
is a little program called sciam that contains the basic kernel that
was published in the September 1992 issue of Scientific American.
To be added to the mailing list, send mail to xerion-request@ai.toronto.edu.
Bugs should be reported to xerion-bugs@ai.toronto.edu. Complaints,
suggestions or comments may be sent to xerion@ai.toronto.edu.
SNNS (Stuttgart Neural Network Simulator) is a software simulator for
neural networks on Unix workstations developed at the Institute for
Parallel and Distributed High Performance Systems (IPVR) at the
University of Stuttgart. The SNNS simulator contains a simultor kernel
written in ANSI C and a 2D/3D graphical user interface running under
X11R4/X11R5. It runs under Sun Sparc (SLC, ELC, SS2, GX, GS), DECstation
(2100, 3100, 5000/200), IBM RS 6000, HP 9000, and IBM-PC (386/486). SNNS
includes the following learning procedures: backpropagation (online,
batch, with momentum and flat spot elimin., time delay),
counterpropagation, quickprop, backpercolation 1, and generalized radial
basis functions (RBF), RProp, recurrent ART1, ART2 and ARTMAP, Cascade
Correlation and Recurrent Cascade Correlation, Dynamic LVQ, and Time
delay networks (TDNN). (Elman networks and some other network paradigms
have already been implemented but are scheduled for a later release.)
The SNNS simulator can be obtained via anonymous ftp from
ftp.informatik.uni-stuttgart.de:/pub/SNNS/SNNSv2.1.tar.Z [129.69.211.2].
The PostScript version of the user manual can be obtained as file
SNNSv2.1.Manual.ps.Z. To be added to the mailing list, send a message
to listserv@informatik.uni-stuttgart.de with "subscribe snns <Your Full
Name>" in the message body. Submissions may be sent to
snns@informatik.uni-stuttgart.de. For further information, contact
Andreas Zell, <zell@informatik.uni-stuttgart.de>.
NEOCOGNITRON SIMULATOR
The Neocognitron Simulator is written in C and is available by
anonymous ftp from
tamsun.tamu.edu:/pub/neocognitron.tar.Z [128.194.15.32]
unix.hensa.ac.uk:/pub/uunet/pub/ai/neural/neocognitron.tar.Z
[129.12.21.7]
PLANET (aka SunNet)
Simulator that runs under X Windows. Written by Yoshiro Miyata
<miyata@sccs.chukyo-u.ac.jp> of Chukyo University, Japan.
Available by anonymous ftp from
tutserver.tut.ac.jp:/pub/misc/PlaNet5.7.tar.Z [133.15.64.6]
boulder.colorado.edu:/pub/generic-sources/PlaNet5.7.tar.Z [128.138.240.1]
Includes documentation.
LVQ_PAK and SOM_PAK
LVQ_PAK (Learning Vector Quantization) and SOM_PAK (Self-Organizing Maps)
were written by the LVQ/SOM Programming Team of the Helsinki
University of Technology, Laboratory of Computer and Information
Science, Rakentajanaukio 2 C, SF-02150 Espoo, FINLAND. The PAKs
run in Unix and MS-DOS systems. Available by anonymous ftp from
cochlea.hut.fi:/pub/lvq_pak/ [130.233.168.48]
cochlea.hut.fi:/pub/som_pak/
ToolDiag
ToolDiag is a feature selection program that increases the accuracy of
classifiers and reduces their complexity by providing them with a
subset containing only the most relevant features. It has interfaces
to LVQ_PAK and SNNS, and uses a data file format that is compatible
with that of LVQ_PAK. The 2-d graphics can be displayed using the
GNUPLOT plotting package. ToolDiag implements many concepts from
Devijver and Kittler's book "Pattern Recognition -- A Statistical
Approach" (Prentice Hall, 1982), including the optimal branch and
bound search strategy, together with several different selection
criteria. ToolDiag can also perform an error estimation using the
leave-one-out method and a K-nearest-neighbor classifier. It also
includes a learning module (Q*) that has the same functionality as
LVQ. ToolDiag cannot handle missing values and requires continuous or
ordered discrete numerical features. ToolDiag is implemented in C and
documentation and source code are available by anonymous ftp from
ftp.fct.unl.pt:/pub/di/packages
For more information, contact Thomas Rauber <tr@fct.unl.pt>.
MACTIVATION
ftp.cs.colorado.edu:/pub/cs/misc/ [128.138.243.151]
as the file Mactivation-3.3.sea.hqx.
DartNet
A Macintosh-based Neural Network Simulator with a nice graphical
interface. Available by anonymous ftp from
dartvax.dartmouth.edu:/pub/mac/dartnet.sit.hqx [129.170.16.4]
or by email from bharucha@dartmouth.edu. New network architectures
and learning algorithms can be added to the system by writing small
XCMD-like CODE resources called nDEF's ("Network Definitions"). For
more information, send email to Sean P. Nolan,
<sean@coos.dartmouth.edu>. [As of 7/7/93, email bounced.]
NevProp is a C implementation of general purpose backpropagation
software, based on Quickprop 1.0 by Scott Fahlman, as translated from
Common Lisp into C by Terry Regier. It runs on Unix, Macintosh, and
DOS. The quickprop algorithm itself has not changed substantially, but
it now includes options to force gradient descent (per-epoch or
per-pattern), generalization & stopped training, c index, and interface
enhancements. It is available by anonymous ftp from
unssun.scs.unr.edu:/pub/goodman/nevpropdir/ [134.197.10.128]
as the file npxxx.shar (replace xxx with the version number) or
from the CMU Simulator Collection. For further information, contact
Phil Goodman <goodman@unr.edu>.
TCS (Tasmanian Connectionist Simulator) is a neural network
simulation package written in Borland C++ for Windows available by
anonymous ftp from
ftp.psychol.utas.edu.au:/pub/tcs [131.217.35.98]
For further information, write to Zoltan Schreter Dept. Psychology
University of Tasmania Hobart, Tasmania AUSTRALIA,
<zoltan@psychnet.psychol.utas.edu.au>.
The HYPERPLANE ANIMATOR is a program that allows convenient graphical
display of the training data and weights in a back-propagation neural
network. As learning progresses and the weights in a neural net
alter, the hyperplane positions move. At the end of the training they
are in positions that roughly divide training data into partitions,
each of which contains only one class of data. Observations of
hyperplane movement can yield valuable insights into neural network
learning. The Animator, developed by Lori Pratt and Steve Nicodemus
of the Colorado School of Mines, uses the Motif toolkit on an IBM
RS6000 with X-Windows. The system currently animates only hyperplanes
representing input-to-hidden weights. The animator is available by
anonymous ftp from
mines.colorado.edu:/pub/software/hyperplane-animator/ [138.67.1.3]
as the file hyperplane-animator.tar. An openwindows version of the
animator is available by anonymous ftp from
cs.rutgers.edu:/pub/hyperplane.animator
For more information, write to lpratt@mines.colorado.edu.
SUZY is a simple neural net classifier system for PCs written in C++
and Turbo Vision. RBFs are used to implement the classifier system
with a class-based algorithm being applied to find the centres and
radii of the RBS units. The program is not intended for any serious
applications and is quite slow, but may be of interest to some people.
It is available by anonymous ftp from
rhino.cis.vutbr.cz:/pub/software/ai/suzy.tar.Z [147.229.3.10]
For further information, contact tgrove@psycho.fme.vutbr.cz.
MBP (Matrix Back Propagation) is an efficient implementation of the
back-propagation algorithm for current-generation workstations. The
algorithm includes a per-epoch adaptive technique for gradient
descent. All the computations are done through matrix multiplications
and make use of highly optimized C code. The goal is to reach almost
peak-performances on RISCs with superscalar capabilities and fast
caches. On some machines (and with large networks) a 30-40x speed-up
can be measured respect to conventional implementations.
The software is available by anonymous ftp from
risc6000.dibe.unige.it:/pub/ [130.251.89.154]
as MBPv1.1.tar.Z (unix version) and MBPv11.zip (DOS version). The
documentation is included in the distribution as the postscript file
mbpv11.ps. For more information, contact Davide Anguita
<anguita@dibe.unige.it> or <anguita@icsi.berkeley.edu>.
THE BRAIN is a neural network (backpropagation) simulator for MSDOS
systems. It is simple enough to be used by non-technical people,
yet sophisticated enough for serious research work. It is available
by anonymous ftp from
ftp.technion.ac.il:/pub/unsupported/dos/local/ [132.68.1.10]
ftp.tu.clausthal.de:/pub/msdos/misc/ [139.174.2.10]
as the file brain12.zip. For more information, write to
David Perkovic <dpc@mep.com> or <perkovic@cleese.apana.org.au>.
PO Box 712, Noarlunga Center SA 5168, Australia.
Neural Systems (Biological Simulation):
BIOSIM is a biologically-oriented neural network simulator. It
implements four neuron models: a simple model only switching ion
channels on and off, the original Hodgkin-Huxley model, the SWIM model
(a modified HH model) and the Golowasch-Buchholz model (the most
enhanced model). Dendrites consist of a chain of segments without
bifurcation. It is in the public domain and runs on Unix workstations
(a less-powerful PC version is also available). BIOSIM includes a
graphical user interface and was designed for research and teaching.
It is available by anonymous ftp from
ftp.uni-kl.de:/pub/bio/neurobio [131.246.9.95]
For more information, write to Stefan Bergdoll <bergdoll@zxa.basf-ag.de>.
GENESIS (GEneral NEural SImulation System) is a general purpose
simulation platform which supports the simulation of neural systems
ranging from complex models of single neurons to simulations of large
networks made up of more abstract neuronal components. Most current
GENESIS applications involve realistic simulations of biological
neural systems. Although the software can also model more abstract
networks, other simulators are more suitable for backpropagation and
similar connectionist modeling. GENESIS and its graphical front-end
XODUS are written in C and run on SUN and DEC graphics work stations
under UNIX (Sun version 4.0 and up, Ultrix 3.1, 4.0 and up), and
X-windows (versions X11R3, X11R4, and X11R5). The current version of
GENESIS has also been used with Silicon Graphics (Irix 4.0.1 and up)
and the HP 700 series (HPUX). The distribution includes full source
code and documentation for both GENESIS and XODUS as well as fourteen
demonstration and tutorial simulations. Documentation for these
simulations is included, along with three papers that describe the
general organization of the simulator. GENESIS is available by
anonymous ftp from genesis.cns.caltech.edu (131.215.137.64). Before
using ftp, you must telnet to genesis.cns.caltech.edu and login as the
user "genesis" (no password required) to register. If you answer all
the questions asked of you an 'ftp' account will automatically be
created for you. You can then 'ftp' back to the machine and download
the software. Further inquiries concerning GENESIS may be addressed
to genesis@cns.caltech.edu.
Probabilistic Reasoning:
BELIEF is a Common Lisp implementation of the Dempster and Kong fusion
and propagation algorithm for Graphical Belief Function Models and the
Lauritzen and Spiegelhalter algorithm for Graphical Probabilistic
Models. It includes code for manipulating graphical belief models such
as Bayes Nets and Relevance Diagrams (a subset of Influence Diagrams)
using both belief functions and probabilities as basic representations
of uncertainty. It is available by anonymous ftp from
ftp.stat.washington.edu [128.95.17.34]
and by email from the author, Russell Almond <almond@stat.washington.edu>.
Contact the author at almond@statsci.com for information about a
commercial version GRAPHICAL-BELIEF currently in the prototype stages.
IDEAL is a LISP system developed for building and evaluating influence
diagrams and Bayesian networks. It is accompanied with a graphical
user interface (CLIM-based) for constructing, editing, and solving
belief networks and influence diagrams. For more information, write
to srinivas@rpal.rockwell.com.
Planning:
NONLIN -- cs.umd.edu:/pub/nonlin (128.8.128.8)
Contact: nonlin-users-request@cs.umd.edu
nonlin-bugs@cs.umd.edu
AbTweak is a complete hierarchical, non-linear planner that extends
David Chapman's (MIT 1986) "Tweak" planner as described by
Yang (Waterloo) and Tenenberg (Rochester) in 1989. This implementation
by Steven Woods (1991 Masters Thesis) includes a complete search
strategy suited to abstraction hierarchies known as LEFT-WEDGE (Woods 1991).
This planner and related work predates that of SNLP. AbTweak has a
WWW homepage containing source & related papers accessible on
http://logos.uwaterloo.ca/sgwoods/
AbTweak is also available by anonymous FTP from
logos.uwaterloo.ca:/pub/abtweak/
For more information send mail to Qiang Yang <qyang@logos.uwaterloo.ca>.
RHETORICAL -- ftp.cs.rochester.edu:/pub/packages/knowledge-tools
Contact: Brad Miller <miller@cs.rochester.edu>
SNLP -- cs.washington.edu:/pub/snlp.tar.Z
Contact: weld@cs.washington.edu
Nonlinear planner.
IDM -- sauquoit.gsfc.nasa.gov (128.183.101.29)
Contact: idm-users@chelmsford.gsfc.nasa.gov
STRIPS-like planning.
PRODIGY -- Contact: prodigy@cs.cmu.edu
Integrated Planning and Learning System
SOAR -- ftp.cs.cmu.edu:
/afs/cs.cmu.edu/project/soar/public/Soar5/ -- Lisp Version
/afs/cs.cmu.edu/project/soar/public/Soar6/ -- C Version
Contact: soar-request@cs.cmu.edu
Integrated Agent Architecture.
Supports learning through chunking.
----------------------------------------------------------------
;;; *EOF*
Go Back Up
Go To Previous