NeuroDataDesign / orange-panda-f16s17

Automated EEG data analysis.

5 stars 5 forks source link

February 5th deliverables #50

Closed rmarren1 closed 7 years ago

rmarren1 commented 7 years ago

Commit

Status	Task	Deliverable
:panda_face:	Optimize python abstract discriminibility function (O(n^2) rather than O(n^3)), ~pip~ easy installable	.ipybn pip page @rmarren1
:panda_face:	Interpolation benchmarking using correlation	.ipybn @rmarren1
🐼 :	Benchmark Robust PCA methods	.ipybn @tsunwong123
:panda_face:	MEDA for electrodes (with notes on how to determine what is 'bad')	.csv, HTMLViewer, dir of HTML @nkumarcc

rmarren1 commented 7 years ago

@jovo

ebridge2 commented 7 years ago

@rmarren1 if citing discriminability for functions, should use names that the paper uses. ie, rdf (reliability density function) not partial discriminability.

ebridge2 commented 7 years ago

look at using kullback leibler divergence. basically, fft per electrode, convert to amplitude/power/etc spectrum in Fourier domain, KLD btwn each electrode's spectrum (gives nxn matrix for n electrodes per sub), then directly use distance or something btwn matrices then discriminability computation. I have a friend example somewhere; let me know if interested and I can find it?