A busy month...

It's been a very busy month, and I'm finally posting all the things that need posting. First off, my website now has a Professional Activities section, listing things I've organized, including the upcoming special issue on Human-Level Intelligence for IEEE Intelligent Systems (deadline January 5th!). This also lists the two events I just ran, the Naturally-Inspired AI symposium, and the Spatial Computing Workshop. Both went very well, and have budding communities that must be nurtured.

I've also posted all the papers and talks from the last month:

• Engineered Robustness by Controlled Hallucination, Jacob Beal and Gerald Jay Sussman, AAAI 2008 Fall Symposium "Naturally-Inspired Artificial Intelligence", November 2008.
  
• Cells Are Plausible Targets for High-Level Spatial Languages, Jacob Beal and Jonathan Bachrach, Spatial Computing Workshop, October 2008. (and talk)
  
• Empirical Characterization of Discretization Error in Gradient-based Algorithms, Jonathan Bachrach, Jacob Beal, Joshua Horowitz, and Dany Qumsiyeh, IEEE SASO 2008, October 2008. (and talk)
  
• Cognitive Security for Personal Devices, Rachel Greenstadt and Jacob Beal, First ACM workshop on AISec, October 2008.
  

Whew.

Proto tutorial slides posted

First of a batch of updates about to go up, getting done and promoted first due to urgency: I've now posted the slides for the Spatial Computing tutorial that we gave at this year's SASO conference, along with the accompanying live demo code.

I'm also listing them on the Proto downloads site as a stop-gap measure until I get the proper tutorial written.

MIT Proto released more than a month ago

Somehow, until today, I managed to forget to announce our release of MIT Proto. It's now posted on my web site.

Milestones

I have now started my job as a Scientist at BBN Technologies. So far, so good, and I'm glad I chose them.

ICDL 2008

I'm just back from presenting Analyzing Composability in a Sparse Encoding Model of Memorization and Association, a paper I wrote with Tom Knight on applying digital circuit design principles to debug a sparse neural coding model.

The International Conference on Development and Learning, where I presented the paper, is a fairly young but high quality conference that's filling in one of the gaps between AAAI and CogSci. I had a blast, and ICDL also has the distinction of being the conference I've attended with the *highest* percentage of people that had something to say to me, and vice versa.

My paper at CogSci

I've been at CogSci the past few days, where I have a paper on Learning Composable Signals for a Cognitive Substrate. It's an extract of the top-level experiments from my dissertation, cast in the larger framework of the cognitive substrate hypothesis, which holds that human intelligence can be explained in terms of a collection of relatively simple specialists that integrate together to form a unified mind.

Congraduations to Dany Qumsiyeh!

I've just graduated my first student: Dany Qumsiyeh turned in an excellent M.Eng. thesis yesterday called A Distributed Building Evacuation System.

Talks, talks, talks

I went to post the Autonomy in Spatial Computing talk I gave at HotAC yesterday and discovered that I'd forgotten to posted the talk on Spatial Computing and the Challenge of Engineered Emergence that I gave at Harvard back in April.

Also, flying for a single-day trip is a bizarre experience. First, it's totally weird to be getting on a plane with nothing but a laptop bag for luggage. Second, the self-service checkin automatically checked me in for my flight back too, meaning that when I got to the airport on the way back, all I had to do was walk straight through security to my gate.

Focus & Autonomy

Two more short papers up: one on Shared Focus of Attention for Heterogeneous Agents that I presented as a poster at AAMAS two weeks ago, and one on Autonomy in Spatial Computing for the Hot Topics in Autonomic Computing workshop next week. A couple more are in the pipeline and will be posted after I present them...

A blast from the past...

I just discovered a *very* old paper online... The Development of a Small Autonomous Helicopter Robot for Search and Rescue in Hostile Environments, a paper from way, way, back in my Junior year of undergrad, when I was a team member of the MIT Aerial Robotics team, trying to get a small (6' rotor span) helicopter to fly on its own. Man, that really takes me back. It's such a happy and positive paper and we never quite got the damned thing to work, because everybody who really understood Kalman filters got recruited into more serious research projects leaving folks like me to tweak the filters and pray. In the end, though, what killed us at the competition was the heat of mid-summer Eastern Washington. The temperature was around 90, and much higher than that a few inched above the blacktop of the competition area, and the nylon gears driving the tail rotor softened and let go, dropping our helicopter from 20 feet up during warm-up trials; fortunately it was still under control of our (amazing) human RC pilot, who brought it in for a controlled crash and our electronics housings protected everything important. Ah well, that's nearly a decade ago now.

A best paper and a position paper.

Our paper on Fast Self-Healing Gradients won Best Paper (Artificial Intelligence & Agents Theme) at the ACM Symposium on Applied Computing. This feels really good because it's a big conference (nearly 400 papers all told) and we were one of six selected for awards. The accompanying talk was very well received. I also felt proud of our Self-Organizing Pervasive Distributed Systems track, because every one of us gave a live demo as part of the talk!

I've also posted a position paper on Cognitive Security for Personal Devices, which I wrote with Rachel Greenstadt, a postdoc at Harvard. We're exploring together some ideas about "trusted computing" machines that own themselves, and how broad-spectrum passive biometrics could allow such a machine to imprint on its owner and make security decisions in a way that is more intuitive to humans.

Synthetic Biology talk

I've posted a talk on programming cell aggregates that I gave to the MIT synthetic biology lunch. This talk gives a brief overview of the Proto approach to spatial computing, then details the set of engineering challenges that must be considered if we are to program organisms at the level of colonies and tissues, rather than individual cells.

Also, I've cleaned up the navigation on my page.