Thanks :) The next part should already be up as part of the Distributed Algorithms CDT channel?
@iidtxbc
2 жыл бұрын
Thanks for the video and hope you can do the next series!
@petergreen5900
2 жыл бұрын
Thanks :) Getting there, just a case of finding the time ...
@gordonjuliankohn3735
Жыл бұрын
Great continuation in the series - please keep going - this introduction is really missing out there!
@petergreen5900
Жыл бұрын
Thank you! Yes I need to do some more, just got to find the time...
@gordonjuliankohn3735
Жыл бұрын
@@petergreen5900 Appreciate your time spent! - it takes time to make it this clear. - In the meanwhile, do you have any reading recommendations on where to pick it up from here?
@petergreen5900
Жыл бұрын
@@gordonjuliankohn3735we had a go at explaining what's going on in this paper: www.sciencedirect.com/science/article/pii/S0888327021004222 but ultimately I think you will end up at the original: academic.oup.com/jrsssb/article/68/3/411/7110641
@gordonjuliankohn3735
Жыл бұрын
@@petergreen5900 Alright, thanks for the references - much appreciated!
@royapar7050
Жыл бұрын
Thank you for the pure explanation. Is q always normal distribution?
@distributedalgorithmscdt6203
10 ай бұрын
No q does not have to be a normal distribution (though I've kept it that way to make things a bit easier here). There are many other, more sophisticated, proposals out there!
@dblueeye
11 ай бұрын
Why do the weights depend on the arbitrary choice of L? More specifically, if we generate a set of samples by some pre-determined K's, the expectation value (of some test function) wpt the final distribution seems to depend on the arbitrary choice of L -- does it make sense?
@distributedalgorithmscdt6203
10 ай бұрын
Hello, L is part of the target distribution over x_1,...,x_K so the weights depend on it by definition, if I've understood the question correctly!
Пікірлер: 13