Homepage of Andrew J. Dougherty

Andrew J. Dougherty
Email: andrewdo@frdcsa.org
Phone: (630) 907-3017
FRDCSA (Formalized Research Database: Cluster Study & Apply)
Aurora, IL

My personal Wiki

Bio

Dia duit, mise Aindílis. Tá mé ag foghlaim Gaeilge anois (ach ar ndóigh is mó chuid Gaeilge go dona), i Aurora, Illinois i Meiriceá. Is mó chroí Gaeilge, Emacs, Debian agus Perl. Níltear sásta anseo mar níl Foclóir Bhearla-Gaeilge chun Emacs go maith aon ait ar chor ar bith! Cén fáth, níl a fhios agam. I gconai tá sé WinGléacht ann. Bhuel, bíonn mé ag leigh neart mo chuid leabhar san leabharlann. B'feideir beidh muid ag cur Robotic Book Scanner dó leabharai san leabharlann.

Andrew Dougherty is an autodidact mathematician and computer scientist specializing in Artificial Intelligence (AI) and Algorithmic Information Theory (AIT) . He is the founder of the FRDCSA project, a practical attempt at weak AI aimed primarily at collecting and interrelating existing software with theoretical motivation from AIT. He has made over 90 open source applications, 400 (unofficial) Debian GNU/Linux packages and 800 Perl5 modules (see http://frdcsa.org/frdcsa).

Results in AIT suggest that a system size is a good heuristic for selecting more capable AI systems (it is a necessary but insufficient condition). The easiest practical, apparently underused method of building AI systems that satisfy this heuristic is to collect, package and interrelate existing AI systems. The proof that this method is not being adopted is the lack of packages for major Linux distributions of extant Linux AI systems such as OpenCyc, ConceptNet, Enju, etc. In fact, most major research AI systems have not been packaged for any operating system that has a package archive. This single limitation is problematic enough to warrant that the major mission of the FRDCSA is to create these packages. So far it has made 400 packages of previously unpackaged AI systems, most of which are now available for public download. The Packager application is a program that semi-automatically generates packages of existing programs, and has completely automatically generated most of the packages in the archive.