[FRIAM] trolling, 'hidden' to 'touch' and 'contact'

[Frank Wimberly]

Re:  machine learning and explanation.
Steve,

If you don't have enough reading to do I recommend "Causation, Prediction,
and Search" by Glymour, Spirtes, and Scheines.  Implementation of
algorithms based on their work was my last job before moving to New
Mexico.  Those machine learning algorithms were focused on inference of
causal relationships based on observational data.  It is often said that
correlation is not causation but those algorithms used conditional
Independence relationships, which are a form of correlation, to infer
causal links between variables in fields like medicine, economics, social
sciences and so on.  In other words, they learned causal graphs
(explanations) in the form of acyclic digraphs given datasets derived from
observations of sets of variables over many cases.

Frank
---
Frank C. Wimberly
505 670-9918
Santa Fe, NM

On Thu, Mar 5, 2020, 9:11 PM Steven A Smith <sasmyth at swcp.com> wrote:

>
> Perhaps coincidentally ... or maybe cause I'm triggered ...:
>
>   Do Atoms Ever Touch?
>   https://youtu.be/P0TNJrTlbBQ
>
> They go 'round and 'round about the definitions and *finally* arrive at the conclusion that "the analogy breaks down". So, the answer to "are emotional states hidden" is "no", but not because they are or are not hidden (by whatever definition of "hidden" you may choose), but because the question is NONSENSE! >8^) So, for all you people who think metaphor is so fundamental ... does this confirm or contradict your bias?
>
> Metaphorist trolling much <grin>?
>
> It confirms *my* bias, but I think your conception of "metaphor as
> fundamental" might be different than my own.  I also may be guilty of
> "moving the goalposts".  I've been studying Category Theory for possible
> formalisms suitable for a sort of universal abstraction of structure
> mappings between different formal structures (e.g. sets, topological
> spaces, vector spaces, posets, manifolds) and the kinds of structures found
> in complex (layered) metaphors.
>
> Just like "scientific theories" or "models" or "maps",  *metaphors are
> always wrong*, some alternately more/less useful/wrong than others.   The
> relation "apt" comes to mind.
>
> I would NOT claim that reality is structured by
> metaphors/analogies/ontologies/models/theories, but rather that our
> *language* and formal understanding is structured in that way.
>
> This maybe ties in to the parallel thread which Nick so aptly dubbed
> "experience beyond experience".  To the extent that there is no way to
> structure what we "know", tying it to what we have "sensed", we have
> "beyond experience"?
>
> It also maybe ties into deep machine learning.  IMO machine learning
> excels at predictive power while being virtually void of explanatory
> power.  The holy grail of neural nets/machine learning/learning classifier
> systems is to analyze the artifact resulting from training well enough to
> provide explanatory power.
>
> FWIW, a few of the resources I've found that provide interesting (far from
> conclusive) insight into all of this are:
>
>    1. Big Picture suggesting how/why CT might inform or relate to MT-
>    https://www.math3ma.com/blog/what-is-category-theory-anyway
>    2. Not very deep or even convincing but broadly referential to the
>    topic -
>    http://bohemiantheory.blogspot.com/2014/03/metaphorically-category-theory-is-study.html
>    3. Meaning, Metaphors, and Morphisms: Theory of Intermediate Natural
>    Transformations - https://arxiv.org/pdf/1801.10542.pdf
>
> I don't know that anyone else here is motivated to dig into any of these,
> but some parallax would be helpful.  The last one is the most
> formal/thorough/promising but does have the added challenge of a (rough)
> translation from Japanese.
>
> - Steve
>
>
>
> ============================================================
> 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
> archives back to 2003: http://friam.471366.n2.nabble.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/20200305/84fa0af0/attachment.html>