Stuff After the Paper
- One of the reasons I'm doing this stuff is that it works with thesis
students.
- Over the summer, I had a couple working on this stuff with the
iris task.
- One thing that was good is that the mechanism seemed to work with
all of the standard STDP models (about 6).
- Another was that you could train multiple times (through the
75 inputs), with a lower learning rate, and it still worked.
- A third was that there was performance variance with different
time steps.
- None the less, there wasn't really much in the way of progress, and
we won't be presenting that here next year.
