Home Reinforcement Learning and Artificial Intelligence (RLAI)

What's new in rlai.net
--Rich Sutton, Sep 12 2004
This page does not seem to be reliably updated; perhaps it is not useful, perhaps it should be replaced with an archive of notifications.  This page will probably be detached soon.  -rich

The ambition of page is to highlight recent significant changes to rlai.net. Anything new and substantive that RLAIers should know about should be listed here, in chronological order (new things at the end).  For changes made after this page was converted to an open page, the author and date of each change is indicated by a signature mouseover (the small bluish square at the end).

30 May 2004 rich, steph
A reasonably stable design for the new python RL interface.
4 Jun 2004 rich, brian
Rich and Brian submitted a paper on TD Networks to NIPS.
14 Jun 2004 rich
This "New Content" page.
14 Jun 2004 rich, satinder
The online proceedings for the ICML'04 workshop on Predictive Representations of World Knowledge.
14 Jun 2004 rich
Documentation for the tile coding software moved to rlai.net. Added a preliminary design for the python version of tile coding.
14 Jun 2004 rich, anna, priyanka
4 new pages of design documents for the Next Gen Web Site.
16 Jun 2004 rich
Draft description of the proposed Python version of the G graphics language.
22 Jun 2004 rich
Topics that Rich sees as excellent research opportunities and essential to solving AI
20 July 2004 rich
The RL Toolkit timeline updated.
03 August 2004 steph
New version of g, graph, graph3d G uploaded
26 August 2004 steph
Graph G documentation and new version added
2 September 2004 steph
New versions of RLAI and RLtoolkit RLtoolkit added
This page converted to an open web page. 
NIPS workshop page "Reinforcement Learning: Benchmarks and Bake-offs" created 
RL interface documentation updated; RLtoolkit pages started. 
RLAI utilities updated with bug fixes for the graph routine. 
Sept 28, 2004: RLAI (graphic) utilities have been merged into the RLtoolkit. The wrap version of tiles and a more efficient naive traces have also been added. Documentation for installing the toolkit and for graph have been updated. 
Top-level page (rlai.net) converted to an open page (/rlai.html). 
Oct 20: New version of RL toolkit added. Changes to demos, new top level demos function; faster Gridworld. 
A collection of scientific hypotheses related to RLAI: reward hypothesis, value-function hypothesis, empirical knowledge hypothesis, and more, and more to come.   -- Rich Sutton Sept 10 2004

Collection of web pages on reinforcement learning in the game of Go.  --David Silver Sept 2004

New web pages added on the robot-rights debate.  -- Rich Sutton, Thu Oct 13 2004

Rich's topics revised/augmented/updated.   -- Rich Sutton, Thu Oct 21 2004 
New pages added on programming style recommendations.

A new top-level G page.  See here.  Also revisions to the top of the page on the python version of G.
Nov 11: Start of software development pages added. Also a new "pre-beta" version of the toolkit.  
This page does not seem to be reliably updated; perhaps it is not useful, perhaps it should be replaced with an archive of notifications. This page will probably be detached soon. -rich 
Extend this Page   How to edit   Style   Subscribe   Notify   Suggest   Help   This open web page hosted at the University of Alberta.   Terms of use  2918/1