Friday, June 19, 2009

Back From Atlanta

I'm back from the IJCNN in Atlanta, where I presented my paper "Sequential Hierarchical Recruitment Learning in a Network of Spiking Neurons". Sounds like a barrel of monkeys, don't it?

Even though I presented on the last day, attendance to the session on spiking neural networks was good. Eugene Izhikevich was in the audience, but didn't say or react much to the talks. Incidentally, his talk on large-scale brain models was very nice. I've been increasingly skeptical about the approach of trying to make enormous models when we have such little grasp of how small, local circuits in the brain work, but he made a very good case. I see the usefulness of large-scale models for studying global phenomena and simply have available a model of that magnitude to tweak and study. Hopefully the large-scale and small-scale models will one day be able to tie all the theory together in one, nice coherent bundle.

John Hopfield's talk was also a highlight. The theme was basically that you want to pick hardware that's best going to fit with the type of algorithm you need to run, and that evolution leads to such efficient coupling. Thus, if we want to try to understand the algorithms of the brain, we need to pay close attention to the type of operations that neurons carry out very well. His conclusion was that understanding the synchronous operations of populations of neurons is key to understanding how they learn and process information. I wholeheartedly agree. :)

Another highlight was a 3-hour tour of some neuroscience labs at Emory. I got to see live recordings from the network that controls the involuntary "swallowing" in crabs and lobsters. I got to see how they make brain slices from rats and mice (first you drug, then decapitate the animals, then you use a razor blade affixed in a machine that's moving back and forth very fast). Another group was studying a group of neurons in leeches which control their heartbeat. Another group was monitoring cells in awake, alert mice, studying how cells in their auditory cortex respond differently to sounds of mice pups depending on whether or not they have given birth to them. And yet another group was monitoring the activity of cells in a rat's hippocampus as it explored novel objects.

Just as with the Boston conference, I'm exhausted, though. Lots of information to assimilate, so time to fall into bed and sleep.

1 comment:

Ian said...

Wow, sounds like quite the conference. Being in the same room as Hopfield would make my decade...