Measures

[dbloom2]

1. Voltage amplitude variability
2. Not Union (Tonic, Quiescence, and Bursting)

[hinsley]

1. Spike voltage amplitude variability: 9927482

I am not yet sure how to handle 2. I have a method for differentiating quiescence, tonic spiking, and bursting from one another, but it's not amenable to finding behavior that doesn't fall neatly into any of the three categories. I believe that inter-spike interval variance/uniformity might be a decent measure to capture what is intended by this, as it can capture edge cases that don't fit neatly into one of those three categories.

ISI variance is easy enough to measure, but is susceptible to being really high with long inter-burst intervals, so will potentially give a lot of noise in scans. ISI uniformity would be a better option, but this requires something like persistent entropy to implement, which would take much longer to get working.

[hinsley]

Inter-spike interval variance: 3904c60

Closing until we find other measures to implement.

[hinsley]

Another to implement: conditional block entropy. Per Jack:

function blockentropy(str,m)
    blocks=[str[i:i+m-1] for i in 1:length(str)-m-1]
    psm = [count(==(b),blocks) for b in unique(blocks)]./length(blocks)
    return -sum([p*log(p) for p in psm])
end

Edit: Implemented here: e1234d776dc2ee0e5926d9e00b7dc8d5fc545789. Closing again.