Why 3B uses all coordinates of every iteration for reconstr?

Bayesian analysis of blinking and bleaching

Why 3B uses all coordinates of every iteration for reconstr?

Postby hoha » Wed Jul 30, 2014 4:42 pm

Hi,

We are trying reconstructing some images and we have looked that you take all the coordinates of the result file, no matter the iteration in which they were computed. Our question is that if the model is being refined in every iteration why not just take from the result file the coordinates computed in the latest iteration?


Looking forward...
Haydee O. Hdez. Aviña
Laboratorio Nacional de Microscopía Avanzada
Instituto de Biotecnología
Universidad Nacional Autónoma de México
hoha
 
Posts: 9
Joined: Mon Apr 14, 2014 9:24 pm

Re: Why 3B uses all coordinates of every iteration for recon

Postby edrosten » Fri Aug 15, 2014 12:24 pm

hoha wrote:Our question is that if the model is being refined in every iteration why not just take from the result file the coordinates computed in the latest iteration?


This is essentially because we are using a Bayesian approach rather than a maximum likelihood fit. Our algorithm is a sampling algorithm is the latest iteration is not necessarily better than the previous iteration, it's a different sampling instead. There is a notion of convergence, but this is a probabilistic convergence to drawing points from the same distribution, rather than convergence to a local maximum. It's the MCMC sampling of the fluorophore states which makes the result of the fitting stage noisy.

The trouble is that the data is too noisy and ambiguous to reliably perform a maximum likelihood fit.
edrosten
 
Posts: 42
Joined: Wed Jul 18, 2012 2:54 pm

Re: Why 3B uses all coordinates of every iteration for recon

Postby hareem » Thu Jan 08, 2015 7:25 am

Hi edrosten,

Thank you very much. These information are really helpful.

Thanks.
hareem
 
Posts: 1
Joined: Thu Jan 08, 2015 7:08 am


Return to 3B

Who is online

Users browsing this forum: No registered users and 2 guests

cron