[FRIAM] looking for a word

[Robert Wall]

Glen,

I believe what you are trying to achieve is what we used to call "face
validity."  To achieve accreditation among the domain experts, the model
had to appeal on an empathic level or it was toast. This was not easy to do
at the program level (DAG?) but easier to do a higher level of abstracted
modeling (organic level).  Beyond that, the model had to produce outcomes
(verification) that the model was not, say just overfitting the data.

So it seems you are saying that you are trying to convey an organic feel
from a mechanistic process. With Herny Markam's Blue Brain project, for
example, I think they were doing this same thing by starting with a digital
reconstruction of recognizable parts of a mouse brain's neocortex.
Analogously, I think, you are starting with a digital reconstruction of a
rat's liver lobules.

So the wordsmith challenge is to describe how the mechanistic structure
overlay one-for-one on to the organic structure.  I would think that this
relationship must be functional.  Not sure.

This understanding doesn't "answer the mail" for you but it might help with
the wordsmithing.

Robert

On Fri, Aug 17, 2018 at 4:28 PM uǝlƃ ☣ <gepropella at gmail.com> wrote:

> Interesting.  Robert's mention of "fractally-associative" was attractive
> to me and seems similar to your [dis]assortativity.  But I'm too ignorant
> (so far) to know whether that has any heuristic power.
>
> I now owe ~4 pints, but only have any confidence I'll have to pay up on 1.
> 8^)
>
> Here's the context.  In our *analogy* from our computational model of the
> liver to a referent liver, we use a directed graph (without degenerate
> cycles) to simulate the lobules in various livers (perfused rat, whole
> animal mouse, etc.).  In that graph, some of the "sinusoidal segments" feed
> into our "central vein".  But they do so in a computationally coherent way
> that is physically incoherent. It's a DAG.  The edges don't actually
> *conduct* the molecules.  It's a magical attachment.  One of my more
> biologically inclined colleagues was trying to analogize to the referent
> liver, which is much more ... "organic" ... whereas our analog is much more
> ... "schematic", if that makes any sense.  My colleague is attempting to
> point out the difference between an actual liver's complex "bed" of flowing
> integration versus our analog's engineered ... "managed" ... "magical" ...
> transference.
>
> Part of my motivation for posting this question, here, is that I'm
> pitching for us to implement a more "space-filling" lobule structure than
> that exhibited by our current DAG.  Although my colleague thinks I'm
> arguing against him, I'm actually trying to bolster his argument that, in
> order to build a *strong* structural analogy (and thereby a strong
> behavioral analogy), we might need a computational structure that is more
> analogous to the referent lobule.
>
> And part of my *rhetoric* requires a relatively catchy word/phrase to use
> to indicate our our current DAG is easily face-falsifiable.
>
>
> On 08/17/2018 01:37 PM, Steven A Smith wrote:
> > Glen -
> >
> > I haven't converged on precisely what you are looking for here...   but
> > am fascinated with the question.
> >
> > My best guess at the general area you are contemplating would involve
> > the graph theoretic idea of a "cluster" and/or imply something about
> > (dis)assortativity.    I think maybe what you are talking about are
> > (collections of) nodes with high local clustering coefficients and I
> > *think* with high assortativity.  If I understand your question, Marcus'
> > suggestions, and the finer points of these graph measures, a typical
> > "hub" in the normal sense would have high disassortativity, or in
> > laymans terms, nodes with high degree would connect more to nodes with
> > low degree, etc.  while what you are looking for might be nodes with
> > (relatively) high degree *and*  high assortativity, or nodes that
> > connect to nodes of similar degree...
> >
> > I know this is far from providing "a word"...  but the resulting phrase
> > might be "an assortative cluster" or "a cluster with high assortativity"?
> >
> > Can you say anything more about the underlying system being modeled?
> > Are you trying to fit this to the known/observed structure or it's
> > function, or one implying the other?
>
>
> --
> ☣ uǝlƃ
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College
> to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
> FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20180817/af231766/attachment-0001.html>