Presentation and Epochs
- We train and test using yeast, a standard data set from the Cal Irvine
machine learning database.
- Each item is a vector of 8 values between 0 and 100, and one category
value.
![](figureAloneFinal.GIF)
- For each input feature we stimulate 10 neurons, and 20 for the
category.
- For each input we train for an epoch. This is 40 cycles of
stimulation then 35 cycles of no stimulation.
- We don't reset during training, but do reset during testing.
- A reset removes all the neurons fatigue and activation.