OpenCogPrime:Tutorial Sessions

From OpenCog
Jump to: navigation, search

These are chat records of past tutorial sessions. For the latest OpenCog tutorials go to Handles On with OpenCog.


Chapter references for the OpenCogPrime:WikiBook.

  1. Sep 10, Chapters 1 & 2, Introduction & Overview of the OCP System [ log ]
  2. Sep 17, Chapters 3 & 4, Atoms & Knowledge Representation [ log ]
  3. Sep 24, Chapter 5, The Combo Language & The Reduct Library (plus digressions) [ log ]
  4. Oct 1, Chapter 6, the "Mind OS" aka the OpenCog Core architecture [ log ]
  5. Oct 8, Chapter 7, Probabilistic Logic Networks and Integrative Inference [ log ]
  6. Oct 15, Chapter 7, Probabilistic Logic Networks and Integrative Inference part 2 [ log ]
  7. Oct 22, Chapter 7, Probabilistic Logic Networks part 3, quantifiers and contexts [ log ]
  8. Oct 30, Chapter 8, Economic Attention Allocation [ log ]
  9. Nov 5, Chapter 9, Embodiment: Perception and Action [ log ]
  10. Nov 19, Chapter 10, Goal-Oriented Cognition [ log ]
  11. Nov 26, Chapters 10 & 11, Goal-Oriented Cognition (continued) & Procedure Execution [ log ]
  12. Dec 4, Chapters 12 & 13, Dimensional Embedding & Probabilistic Evolutionary Learning [ log ]
  13. Dec 11, Chapter 13, Probabilistic Evolutionary Learning [ log ]
  14. Dec 18, Chapter 14, Speculative Concept Formation and Chapter 15, Integrative Procedure and Predicate Learning [ log ]
  15. Jan 7, Chapter 16, Map Encapsulation [ log ]
  16. Jan 14, OpenCog: The Roadmap for 2009 [ log ]
  17. Jan 21


In an effort to help others to better understand the OpenCogPrime design -- and also to gain feedback on aspects of the design and help me work kinks out of the manuscript -- I (Ben Goertzel) have decided to run a series of tutorial sessions on OpenCogPrime during Sep 2008- Jan 2009.

The sessions will be roughly weekly, but I might end up skipping some weeks when I'm traveling (this will be announced in advance, of course).

The idea is that each week there will be a 2-hour-long IRC session, on the #opencog channel at the freenode IRC server, focused on a particular chapter of the OpenCogPrime:WikiBook. We'll proceed through the chapters in sequence until everyone understands everything (yeah, right!).

While the book focuses on conceptual issues, it will not be problematic if the discussion veers onto implementation and software design issues relevant to the book chapters under discussion ... in fact this would be quite desirable.

The format I suggest is as follows. Out of each 2 hour session,

  • the first 15 minutes may be devoted to general OpenCog discussion
  • the next 30 minutes may be devoted to discussion of the wikibook chapter discussed the PREVIOUS week (in case folks have further thoughts or questions based on looking at the book again or thinking about the topics more after the previous week's discussion)
  • the next 75 minutes will be spent discussing the wikibook chapter that is the current week's topic of study

This rough schedule may of course be mutated if it doesn't seem right in practice.

Choosing the right time for this is difficult because everybody's schedule is different. My initial suggestion is Wednesday 9-11 PM US Eastern time. My reasoning is that this is after-work-hours for most US folks, yet not so late as to be objectionable for US East-coasters. (Sorry for the US-centricity, but this is where I'm living at the moment.) It also should work OK for those in the Orient. It seems to suck for Europeans, but I don't see how to please everybody. Also, it is easiest for ME to find time for a 2-hour chat session in my evenings ... during the daytime meetings and phone calls and such tend to get scheduled....

PLN Tutorials?

Some interest has been expressed in having tutorials specifically on Probabilistic Logic Networks, based on the PLN book (to be released by Springer soon). This is not a bad idea but I decided it would be too much tutorial-ing to do that in parallel with the OpenCog tutorials. Maybe the PLN section of the OpenCog tutorials will wind up getting stretched out to a few weeks rather than just one.

Or maybe, once the OpenCogPrime tutorials are done, we can start with PLN tutorials at that point -- it'll be more fun then because the PLN implementation will be ready to play with!!