PORTABLE AI LAB: An Integrated Toolkit of AI Techniques and Tools

The Portable AI Lab is a computing environment containing a collection
of state-of-the-art AI tools, examples, and documentation.  It is
aimed at those involved in AI courses (i.e. in both teaching and
learning) at university level or equivalent.  It has been developed
under Swiss National Research Programme PNR 23 on AI and Robotics by
IDSIA Lugano in collaboration with IFI, University of Zurich and the
Laboratoire d'IA at the EPFL, Lausanne. The system is available free
of charge.

The design of Portable AI Lab is motivated by the conviction that the
acquisition of expertise in AI depends on extensive practical
experience with a broad range of AI problems. Students should also
appreciate that such problems are typically interdisciplinary in that
they often involve more than one AI subdomain.  The system has enough
built-in functionality to enable its users to get such experience
without having to build all the supporting tools from scratch and is
intended to encourage the exploration of subdomains and the
relationships between them. 

The functionality is provided though a number of modules concerned with:

	- Automatic Theorem Proving
	- Natural Language Processing
	- Rule-Based Inference
	- Truth Maintenance Systems
	- Constraint Satisfaction Problems
	- Learning
	- Knowledge Acquisition
	- Genetic Algorithms
	- Neural Networks

Each module comprises a fully implememented and documented
computational kernel over the subdomain in question, together with a
representative set of autonomous demonstrations and running examples.
Unlike many similar systems, a lot of attention has been paid to
providing appropriate graphic visualisations for the presentation of
key algorithms and concepts.  In addition, sufficient documentation is
provided to enable the user to retrieve key literature references,
understand the architecture and specifics of the implementation, and
change the latter if desired. 

The current version of the system is implemented in Common Lisp, CLOS,
and Common Windows, all of which are necessary to create an executable
version of the system from the source code. These are normally
available to work with Allegro Common Lisp version 4.1 as supplied for
SPARC workstations.  The system will run acceptably with 16M of memory
given adequate swap space. For those without Allegro CL, a
runtime-only version of the system is available for SPARC
workstations at nominal cost. 

Further details from Michael Rosner, IDSIA, Corso Elvezia 36, 6900
Lugano, Switzerland, email mike@idsia.ch, fax +41 91 22 89 94.