# HG changeset patch # User Yoshua Bengio # Date 1283993117 14400 # Node ID bc3f7834db83042670503be377b9d033e809d0c1 # Parent a474fabd1f378adad23898afe776635d761eb347 added a comment/question about Type diff -r a474fabd1f37 -r bc3f7834db83 doc/v2_planning/learner.txt --- a/doc/v2_planning/learner.txt Wed Sep 08 16:20:20 2010 -0400 +++ b/doc/v2_planning/learner.txt Wed Sep 08 20:45:17 2010 -0400 @@ -270,6 +270,18 @@ Comments ~~~~~~~~ + +YB asks: it seems to me that what we really need from "Type" is not just +testing that a value is legal, but more practically a function that specifies the +prior distribution for the hyper-parameter, i.e., how to sample from it, +and possibly some representation of it that could be used to infer +a posterior (such as an unnormalized log-density or log-probability). +Having the min and max and default limits us to the uniform distribution, +which may not always be appropriate. For example sometimes we'd like +Gaussian (-infty to infty) or Exponential (0 to infty) or Poisson (non-negative integers). +For that reason, I think that "Type" is not a very good name. +How about "Prior" or "Density" or something like that? + OD asks: (I hope it's ok to leave comments even though I'm not in committee... I'm interested to see how the learner interface is shaping up so I'll be keeping an eye on this file)