How to cross-validate? Take half of your trials and compute the peak location and sorting index. Then plot the traces made from the other half of the trials, using the sorting index from the first. h/t @MatteoCarandini
-
-
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
"now with jet!"

- Još 1 odgovor
Novi razgovor -
-
-
Except literally everyone x-validates sorted data or uses an unsupervised method like SeqNMF to extract sequences. Also if you average over multiple instantiations of your z, won't this pseudo-seq will vanish right off your panel 4? 1/2
-
I hope people x validate but I rarely see automated methods used (seqNMF is fairly recent). You'd be right about averaging is there trial to trial data was random and not random PSTHs
Kraj razgovora
Novi razgovor -
-
- Još 1 odgovor
Novi razgovor -
-
What did you expect? In i.i.d. noise, the peak time location T_i is uniformly distributed across the time axis - and smoothing doesn't change this. When you sort the sample by T_i, peaks obviously fall into a straight line.
-
And because you're using imagesc, you're guaranteeing that this max value is plotted using extreme colour from the colour map.
Kraj razgovora
Novi razgovor -
-
-
"Never let the truth get in the way of a good story"
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.