How to use bayesoptdisc function

[babelpro...@gmail.com]

Hi,

I'd like to use the bayesoptdisc function in Octave to optimise the number of layers in a neural net and the number of nodes in each layer, but I have a question about the xset input for the function. I understand that this must be a matrix, so, for example, if my simple xset matrix was [ 10 40 ; 20 60 ] where the first column is number of nodes in the first layer ( 10 or 20 ) and the second column is the number of nodes in the second layer ( 40 or 60 ), what combinations would the bayesoptdisc function actually evaluate? Would it try to evaluate all four possible combinations or just the two combinations represented by the two rows, i.e. just [ 10 40 ] and [ 20 60 ]. If it is the latter case, then to ensure all four combinations are (possibly) considered, I would have to provide an xset matrix of [ 10 40 ; 10 60 ; 20 40 ; 20 60 ]. Is my understanding correct?

[Ruben Martinez-Cantin]

You have to provide all posible combinations. Think of it as
coordinates of cities in map.

When the points are in a grid, like your case, I know it may look like
an overkill, but the advantage of this method is that allows points
that are not in a grid pattern.

Ruben

> --
> You received this message because you are subscribed to the Google Groups
> "BayesOpt discussion" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to bayesopt-discus...@googlegroups.com.
> To post to this group, send email to bayesopt-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/bayesopt-discussion.

[babelpro...@gmail.com]

Hi,

Thanks for the quick reply.

Just one more quick question then - since I will have to explicitly supply a large matrix with all possible combinations, does the bayesoptdisc function extract column vectors or row vectors from the xset matrix as the vector input to the user supplied function which is to be evaluated? My reading of the demo.m function shows that column vectors are extracted. Is this correct?

> email to bayesopt-discussion+unsub...@googlegroups.com.