This follows up prior AGI Summer Schools in Xiamen (2009) and Rekyjavik (2012), though the format will be a bit different.

Ben Goertzel and Lake Watkins will be running a 2-day session of the Summer School on OpenCog…

Check out details at

http://www.mindmakers.org/projects/agi-summer-school-2013

… we hope to see you there!

• Integration of ROS (Robot Operating System) with OpenCog: Arsalan Akhter, mentored by Matt Chapman
• Packages for Ubuntu (and other Unix-like systems) and Cygwin port: Amber Jain, mentored by Nil Geisweiller and David Hart
• DeSTIN/OpenCog integration: Alexander Morrow, mentored by Ben Goertzel
• Extend MOSES to Encompass Primitive Recursive Functions Using Fold: Mandi Wang, mentored by Linas Vepstas
• Spatial and Temporal Reasoning: Dario Garcia-Gasulla, mentored by Jared Wigmore

We’re all excited to be participating in GSoC again, and to have a great bunch of students working on some important projects…

Work on the OpenCog Hong Kong project is also going well — we’ve brought a full-time game developer (Lake Watkins) onto the project to help build a more interesting game-world and game-play on top of our Unity3D-based “OpenCog teaching/training platform.”  An additional graphic designer has been brought into the team and is working with Lake.  The game world is somewhat Minecraft-like in that many things in the world are made of blocks, although the gameplay is very different from Minecraft, involving a robot, a little girl and a ghost dealing with strange situations in a complex mansion.  Coordinated with this game-world work, Shujing Ke has been busy replacing the 2DSpaceMap in OpenCog with a 3DSpaceMap, which has required large changes in multiple parts of the code, and led to a general cleanup of lots of space-related OpenCog code.  (And the rest of the OpenCog HK team has been doing interesting stuff too, to be reported in later OpenCog updates…)

Also arriving with me that day were:

• Dingjie Wang, a M.S. student at Xiamen University who will soon join the Opencog HK team,
• Oliver “Lake” Watkins, an experienced game designer and programmer, who will be staying in Hong Kong to help develop a video game demo to showcase the work of the OpenCog team here at the M-lab,
• Eddie Monroe, an employee of Novamente from Virginia working on Brain Simulation,
• and myself, a volunteer OpenCog enthusiast from California.

We all had different purposes in coming to Hong Kong, but a shared goal of learning more about OpenCog and contributing to it’s advancement. My personal goal has been to help improve project documentation and develop a series of several brief developer tutorials to help other open-source contributors get up to speed with OpenCog more quickly.

Throughout the week, the four of us met daily with Ben Goertzel, Jared Wigmore, and Shujing Ke for presentations about the various components of OpenCog and their present usage and development in Hong Kong. We focused especially on Moses and the various components being used by the agent in the Unity3D world, including OpenPsi and the Perception Action Interface. We also had a couple creative brain-storming sessions with Lake to discuss video game ideas that could be suited to demonstrating the OpenCog game agent. On Friday, we celebrated Future Day together at an excellent Indian Restaurant. Finally, on Saturday, I got to attend TEDxHKUST where Ben spoke about radical life extension and the role AGI may play in helping us to discover therapies to stop aging.

There were several outcomes during my visit:

• It was decided that the Unity3D world, although still undergoing much development, ought to be open-sourced for other Developers to experiment with. Expect another post to this blog when that code repository has been opened to the public.
• Shujing wrote some initial documentation for setting up the Unity world.
• Shujing also developed a brand new demo to showcase the 3D mapper she’s been working on. It allows the agent to recognize a set of blocks as a single entity. The demo is available on YouTube.
• Various updates and improvements were made to the OpenCog wiki, including the addition of documentation for setting up Gephi to visualize the Atomspace, thanks to Jared’s help.
• With help from Jared, Lake, and Nil Geisweiller (via IRC), I was able to develop a patch for the MOSES executable that allows an option to output the final program as a complete Python module. That patch [6] will hopefully be committed to the main repository soon, pending further review. I’m not sure if this will be useful to other developers, but it was a great learning experience for me.
• We discussed various ways to grow the open source community around OpenCog, and evaluated past experiences with Google Summer of Code. This led to a decision to re-apply the OpenCog Foundation for this year as a mentoring organization.

Although I’m leaving Hong Kong, I hope to continue working with the OpenCog team remotely as I prepare materials for the Community CogCamp to be held in Los Angeles in April. This event is the next step in my plan to bring more Open Source developers into the OpenCog project. In the spirit of BarCamps and Hackathons, it will be gathering of interested software developers ready to share knowledge and work together to learn how to use and extend the tools that OpenCog offers. By that time, I also hope to have the beginnings of a ROS integration layer that will do for physical robots what the Perception Action Interface does now for the Unity3D agent. Day 2 of the CogCamp will be focused on that component.

I’m incredibly grateful to Ben and the team for welcoming me in Hong Kong, and I’m eager to continue to apply my new knowledge toward our shared goal of building thinking machines. I look forward to seeing my new friends again at AGI Conferences around the world, and maybe some of you at the LA CogCamp in April.

Future Day

Future Day is a way of focusing and celebrating the energy that more and more people around the world are directing toward creating a radically better future.

Future Day @Second Life – Terasem Island

This is a public event held at Terasem Island, Second Life.

Note start times:
6:00 PM EST UA (GMT-5 March 1st)
12:00 MIDNIGHT EU (GMT+1 End of March 1st)
7:00 AM HONG KONG (GMT+8 Match 2nd)
10:00 AM AUS (GMT+11 March 2nd)
Timezones: http://bit.ly/y9pTI6

Schedule / Lineup:
1. Natasha introduces event 1-2 minutes
2. Ben Goertzel – 10 minutes
3. Howard Bloom – 10 minutes
4. Martine Rothblatt – 10 minutes
5. Adam A. Ford – 10 minutes
6. Giulio Prisco – 10 minutes
7. Natasha Vita-More – 10 minutes

5:30 for drinks, 6:30 sharp panel start. Graduate House, 220 Leicester Street, Carlton
Join project of Kororoit Symposium, Humanity+ Australia and Singularity Summit AU
Welcome: Adam Ford, Humanity+ Australia and Singularity Summit AU organiser
6:30pm Panel Moderator: Josie Taylor, ABC TV1 730Vic presenter
Alain Moulet, President, Aménagements Vivants, here for Kororoit Symposium: Living Spaces for Change
Cobina Crawford, Swinburne Strategic Foresight MA student who tweets about Occupy Melbourne
Amanda Dodd, Environmental Planner and convenor Friends of Iramoo (Grasslands at VU St. Albans)
Sean McMullen, Governance and IT Continuity at Bureau of Meteorology and author sci/fi and fantasy
8:00pm Skype in Dr Ben Goertzel – CEO of Novamente, founder of Open Cog
8:20pm Future Jam! – Rapid prototyping exercise (extremely participatory)
9:40pm Skype in Natasha Vita-More – designer and Chair of Humanity+
10:00pm Sean McMullen – Presentation
~10:30pmWrap up (Adam Ford and Tony Smith) and Thank You – Lets make the Future!

According to the Constitution, OCF is a democratically run organization, with a Board elected by its membership, where the latter is defined as individuals who have substantially contributed to the OpenCog project, and have asked to become formal members and been approved by the Board.

The founding Board consists of Ben Goertzel, Cassio Pennachin, Nil Geisweiller and Joel Pitt; it was instantiated Jan. 1, 2012, and according to the Constitution, these Board members will be up for re-election in 2015.

If you would like to become a formal member of OpenCog Foundation and participate in its formal governance please email info@opencog.org and say so.

At the moment there is essentially no formal governance occurring in the OCF!  However, if at some point significant funds are raised for OpenCog development, this may be managed by OCF.  Also, many of the OpenCog core classes have copyright OCF; so it seems valuable to have these classes tied to a democratically-run nonprofit organization.

Different people may write the Roundups different months — this one is coming from Ben Goertzel…

As before, the Roundups will largely focus on stuff that’s been completed during the time since the last roundup, with additional quasi-random mentions of work in progress or new ideas as well.  They’re going to be informal and chatty rather than formal progress reports — so enjoy or ignore

What’s been going on in January OpenCog-wise?  Among other things…

Linas Vepstas has been digging deep into the MOSES code (which, for you newbies out there, does probabilistic program evolution).  He’s fixed up the learning of programs involving both discrete and continuous inputs, and is now looking at some possible fundamental improvements to the learning algorithm.  Meanwhile, Nil Geisweiller and I have been using MOSES for some quantitative data analysis applications.

Jared Wigmore has connected his shiny new Python implementation of PLN (Probabilistic Logic Networks) to OpenPsi (which handles motivated action), enabling goal-directed planning to get done using PLN (instead of the simpler crisp planner that was previously being used with OpenPsi).

Some interesting work in progress in the OpenCog Hong Kong group includes: Jared upgrading the his “Frequent SubHypergraph Miner” (Fishgram) to be a “Surprising SubHypergraph Miner” (Sushygram); and Shujing Ke working toward the use of Sushygram for identifying the boundaries of events (in the virtual world where the OpenCog HK projects little OpenCog game agent lives).   The work on event boundary recognition has led us to realize that the recognition of simple spatial events in the virtual world needs more work, and Shujing has also been working on that, with some help from Eddie Monroe (who is in the US).

On the RelEx (language processing side), Ruiting Lian is preparing a corpus of RelEx parses for someone to rate, so that we’ll have a rigorous (albeit not that large) basis for assessing RelEx performance, and the performance of approaches to RelEx parse ranking.

There have also been  many interesting discussions with David Hanson about potentially connecting OpenCog with his Robokind robots and other robots he’s brewing up.

And some sort of near-consensus seems to have been achieved that the OpenCog core-system revamp suggested a couple months ago by Cassio Pennachin and others is a good idea.  So we’re trying to figure out how to fund someone to do that during 2012.

For those who have been asking about the Building Better Minds manuscript, which summarizes the AGI vision that led to the creation of OpenCog in the first place  — I’ve been going through the chapters one by one and improving the introductions and conclusions and some of the innards, to better tie the technical material into the underlying theory.   By late Spring I hope to have a version to send to a dozen or so enthusiasts/experts to scrutinize, and after incorporating their comments I’ll do the final revision…

And so it goes … never as fast as I’d like, but still, moving forward

• bringing newbies up to speed on OpenCog
• collaborating on OpenCog development with others who already have a little experience.

The date will be the weekend of Sat-Sun April 21-22; more details are available at:

http://www.meetup.com/LARoboticsClub/events/47919462/

http://www.meetup.com/LARoboticsClub/events/47919542/

Wide-ranging discussion will be welcome, but the focus will be on hands-on technical work with the OpenCog code.