MNIST Results
- Results for both versions of the system are above chance (10%),
but are quite poor.
- Modifying weights for balancing does best at 32.1%. A student
trying it now got 40%
- Adaptive neurons get 25.5% correct.
- We're not aware of any other spiking systems used for
this, but a translation from an MLP would probably get
around 97%.
- It's just a start, and we're hoping to work on this a bit
more and get significantly better results.
