Open jovo opened 9 years ago
That definitely sounds interesting! Potentially something to add as an operation on the TimeSeries
object. Is this intended for massively univariate or multivariate time series operations? Can you tell us more about the use cases / algorithm / implementation?
massively multivariate :) uses case 1 was: whole brain fMRI use case 2 is about to be: whole fish brain! in brief, the algorithm is a penalized expectation maximization algorithm, which uses fista for the m-step. everything factorizes quite nicely, so i think spark might be ideal or overkill :) i could send you a pre-print describing the algorithm, we hope to post the manuscript to arxiv in a couple weeks, we are just finalizing wordsmithing now....
On Wed, May 13, 2015 at 2:12 PM, Jeremy Freeman notifications@github.com wrote:
That definitely sounds interesting! Potentially something to add as an operation on the TimeSeries object. Is this intended for massively univariate or multivariate time series operations? Can you tell us more about the use cases / algorithm / implementation?
— Reply to this email directly or view it on GitHub https://github.com/thunder-project/thunder/issues/180#issuecomment-101762773 .
the glass is all full: half water, half air. openconnecto.me, jovo.me, office hours https://www.google.com/calendar/embed?src=e2ktu4lrgul8anp8hclrcminp8%40group.calendar.google.com&ctz=America/New_York
we have some new stuff on parallelize high-dimensional KFS, which perhaps we should implement in your framework?