Thursday, June 21, 2007

my own bite-sized piece of Neuroscience

Salut!
I spent my first day in the lab last Monday passing around the 7 grad students in the lab, and talking to them about their projects and what a 2 1/2 month intern (moi) could do working with them. The Computational Neuroscience Lab, and maybe the field in general, can be divided roughly between people who have a physics background and people who have a computer science background; research may be on models of a single spiking electron, or on systems level modeling of a particular behavior.

I was very tempted to start working with those working on single electron models because it had the most in common with my limited background. It came down to deciding to work on the project most clear to me in its scope and motivation.

So, I am testing various algorithms that perform Independent Component Analysis (ICA). An everyday example of ICA is when you focus on one person's voice in a noisy room full of competing voices; our brains are able to distinguish meaningful signals when they are superimposed with various other signals. There are many existing ICA algorithms that do the trick, but they do not perform signal processing in a way that is plausible for neurons, i.e. they neglect the temporal order of the signals. The goal is to compare the performance of several ICA algorithms that take different approaches, including one developed here that takes a biological approach.

I find thinking about modeling a brain function with a computation interesting, but the everyday work is not the same as day-dreaming about it's mysterious description. The challenge of the past week and a half has been to figure out what is important from an immense amount of literature full of technical detail. I get re-directed by talking to my grad-student advisor Claudia. She seems to have mastered reading papers for what is relevant without getting stuck in the details. If I manage to learn that skill from her, I'd will be a very grateful intern.

Jess

No comments: