Changes between Version 11 and Version 12 of UWSummerStatsWorkshop


Ignore:
Timestamp:
Jul 22, 2015, 3:00:42 PM (9 years ago)
Author:
iovercast
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • UWSummerStatsWorkshop

    v11 v12  
    106106 * Sensitivity analysis: Bayes factors can be peculiarly sensitive to the priors in ways you can't expect, so testing different priors could be informative. 
    107107 * Model choice: Don't use flat priors on things that are only present in one model 
    108  
     108* Wednesday AM2 
     109 * How to reduce variance of sampling: reduce variance of function sampled or increase the number of samples taken. 
     110 * minimal relevance sampling: Choice of density of sampling will influence the variance of your monte carlo estimate. 
     111 * That's pretty much what importance sampling is all about is multiplying things by 1, dressed up in a "tricky fashion". 
     112 * '''Importance sampling:''' How to sample wisely for your monte carlo estimates. 
     113  * [http://ib.berkeley.edu/labs/slatkin/eriq/classes/guest_lect/mc_lecture_notes.pdf Thorough lecture notes on this]. 
     114 * "Poor mixing is the evil cousin of reducibility" 
     115 * Metropolis-coupled monte carlo (heated chains) 
     116  * Chains with exponent modifiers 0 < x < 1 
     117 * Simulated annealing 
     118  * Chains with exponent modifiers x > 1 
    109119==== quotes ==== 
    110120* "Out of all the tomorrows we might experience...."