Properties vs Characteristics vs Attributes. Major hair splitting. Properties are usually general. Characteristics can be “case-by-case”. I think “attributes” are equivalent to “characteristics”.

Properties vs Characteristics vs Attributes.
 
Major hair splitting. Properties are usually general. Characteristics can be “case-by-case”.
 
I think “attributes” are equivalent to “characteristics”.
—–
 Properties(Characteristics(Attribute))
=====
  Oh it’s confusing stuff.They’re not only distinct but ALSO synonyms of each other.I’m following a notion I had of an “object free” ontology that defines a something without needing a something.
======
 Closest thing I can think of is stuff like “What kind of plant is this?” databases, where it asks you questions and narrows it down until you get what you want (or not).
=====
 This isn’t it, but there’s “e language” which removes all forms of “to be”, turning language into “action oriented”.
=====
Kaufman’s Eifenforms formalize this idea!
 If Herbert Bernstein wasn’t on sabbatical that year, I’d have taken a course or two back in 1990/1991.But, I don’t think he’s through whom I’d heard first of knots. Now I have to figure it out. [how I knew of Kaufman as a god among thinkers from pre web days]https://www.hampshire.edu/faculty/herbert-j-bernstein
===
Bernstein was Lee Smolin’s Advisor.http://leesmolin.com/about-lee-smolin/academic-cv/
——-
 Smolin is a proponent of loop quantum gravity and has remained cautious about string theory.He wrote:
https://en.wikipedia.org/wiki/The_Trouble_with_PhysicsBut I have to find out HOW I knew of Kaufman. I used to hang around all of the Professors I *didn’t* have, hanging out in their studies and asking loads of questions.
=====
Kenneth Hoffman was a professor there and one of his students was David C. Kelly, another professor and I hung around them both a lot, bugging them with lots of questions all of the time.
 
[I’m gonna find that connection. I’m doing it for myself of course but publically as this is as good as space as any]
 
https://www.genealogy.math.ndsu.nodak.edu/id.php?id=22457
=====
Then there was Jay Garfield, philosophy professor that I also didn’t have but hung around, who just about got DEATH threats from a newspaper article criticising Philosophy depts for not only not including non-Western philosophers in philosophy courses but for REFUSING to even CALL it “Western Philosophy”, and “Let’s just call it what it is, shall we?” [ie – racism]
 
Subsequently went on to make this which is now a reference work.
 
But THAT’S not how I knew of Kauffman either. But I’m REALLY enjoying going through 25+ yr old names and finding out updates.
 
https://www.amazon.com/Oxford-Handbook-World-Philosophy-Handbooks/dp/0199351953/ref=sr_1_1?s=books&ie=UTF8&qid=1474462682&sr=1-1&keywords=oxford+handbook+of+world+philosophy
=======
 ok, had to be in CompSci. Gonna find out WHO… Sorry for all these pings but “thinking out loud” is helping. :)
=====
 [it’s basically an impossible task: I’m trying to locate a book I once picked up in a professor’s study, a professor I didn’t have, who talked with me about chaos, complexity and knot theory, who didn’t actually TEACH those subjects — I don’t think. Just a fellow enthusiast].
======
Ah!!! It was Professor Bernstein that I got that notion from. (how foundational quantum mechanics is)
 
[wasn’t even my professor *but* it was his high school seminar for prospective students, “Quantum Mechanics for the Myriad” intro that he did, the course I wanted:
 
17-18 yr old me REALLY wanted this in 1990 – AND he’s still teaching it. Ah well.
 
Quantum Mechanics For The Myriad
Herbert Bernstein
 
This course will investigate the structure of a powerful intellectual influence of our times, theoretical physics. Using two-state systems including electron spin and photon polarization, we develop the actual quantum theory in its matrix mechanics form. This theory underlies our current understanding of atoms, particles, and virtually all physical processes; it has important philosophical consequences as well.
 
The course has three themes: quantitative approximations to interesting phenomena; formal use of mathematics to describe observations; the philosophical and cultural significance of interpretations of physical theory. Students contact course material in ways parallel to physicists approaching nature. How to formulate questions, including how to make them into solvable puzzles, how to work cooperatively-utilizing both learned and created concepts, and how to master formal reasoning are all learned by experience. Class will meet for one and one-half hours three times a week.
 
 
======
 I don’t know. Certainly could duplicate the mechanics in a rough way. But will it learn the algorithmic processes, or do a “top down” ‘looks like” simulation?Most AI is “looks like” rather than reasoning systems.
======
 https://en.wikipedia.org/wiki/Reasoning_system are far more complicated than what most AI does.
=============================
 No. It doesn’t program itself. There’s both supervised and unsupervised machine learning.Almost all are supervised, at least partially.A detection device is created by humans and parameters for “This Is Human” is created by humans.There’s no proof that any of these systems are in use in CURRENT self-driving cars.But similar human detectors have shown bias towards certain skin pigmentations, most likely due to ignorance of the programmers, designers, business planning teams, testers, etc.
=====
 It does, but it’s far less magical than it seems.Greatest successes are with supervised or reinforced learning rather than unsupervised.https://en.wikipedia.org/wiki/Learning_classifier_system
======
 Oh, there’s unsupervised. In certain areas it’s amazing. Four years ago, I had a neural net try to find my face. After 10,000 tries it got close. But there’s far better algorithms out there than are faster of course.https://www.youtube.com/watch?v=LxQuFUPlkkQ
======
 Oh, I 80%-90% certain it’s ignorance.But — that’s a big problem. If the error isn’t noticed by ANYBODY in a company, that says something significant about the company. Intentional is fixable quickly.Bounding boxes is simpler to implement for motion detection.But bounding boxes usually work by increasing/decreasing brightness in images to find common areas.This basic example I wrote 3 years back using a meme as template shows how bounding boxes work.There’s better ways to do it but this method is computationally cheap and fast so it’s most common at the start of even the fanciest motion detectors out there.Works great with grayscale that average to white.Works awfully with grayscale that average to black.https://www.youtube.com/watch?v=1tKOe6DzKjk
======
 The AI is biased (skewed) by engineers who failed to use a broad population sample.
=====
 3 different software packages.
All show similar bias.
You can draw a “typical” assumption from that.
=====
 Both of these improved datasets are a direct result of her published, peer reviewed study.
======
 Those systems she analyzed are propriety and closed.Scientists have zero access to it.What do you do in science? You work with the objective data you have and can quantify (or qualify but usually quantify).”It’s chance” is always possible when scientists find patterns.Religious people say “It’s God”.
=======
 A pattern was identified using a dataset.
=====
  “significant characteristics” are in the output.The data is “black box”. It is hidden from EVERYBODY’S view, except the companies.The companies, faced with this evidence, CHANGED their data sets.
======
 For a commercial product with proprietary data sets, there is no other option.
=======
 In a utopia, all data is open.
======
What is wrong is withholding the dataset.
 ——-
 The scientist is not wrong for working with available data.
========
 maybe
maybe
maybe
Ask IBM.
Ask Amazon.
Ask Microsoft.Scientist had hypothesis.
Scientist performed testing.
Scientist got results.
Scientist published.IBM.responded.
Microsoft responded.
=======
 Software performed same functions.
Had similar results.
======
4 sets of data.
3 unknown source.
1 known.
1 known tested against 3 unknown.3 unknown produce similar results.Hence, “it is possible” that 3 unknown sets are similar.
=======
  That analogy does not apply.In this case, 1 result showing bias would be a curiousity.
2 results showing bias would be a concern.
3 results showing bias is science.It is akin to 3 separate experiments validating each other but far more interesting than that as they were applied to DIFFERENT programs and yet got similar results.
————————–
 Ok. Her sample was 1200 images, not identical.Already a larger dataset (quantifiably) and more diverse (qualitatively) than your 3 identical dice.
=====
 You didn’t read the article so you do not know about her study.
=====
 You are concluding from ignorance.
=======
No.
Typical was inferred from achieving similar results.
======
 Tired yet conclusive that a failed, invalid study was done that could just as well have been random?
======
 Nobody has yet put forth that argument.It started just there with your statement right then.
======
 Algorithms can be biased, yes. Human programmers do it all of the time. (try being from a Spanish country with a long name designating your lineage and use a database with “First Name / Last Name”)But that’s an easy fix.Machine learning develops biases from the datasets.That’s the whole point of this.
======
 GIGO.
Nobody says that anymore.
But from my early days learning programming:
“Garbage In
Garbage out”.Still holds true.
——————————
 Datasets don’t just sit around doing things.
Algorithms don’t act unless run by a human.
Performance features don’t fall from the heavens.
======
 Computer algorithms are ALL second order logic.Just because a program compiles, doesn’t make it correct.
======
 Even a program written to verify first order logic, is written with second-order logic.
======
 Built on Church/Turing machines. All 2nd order. That’s the basis of computing and what “worked around” Godel.
=
——
 The machine it runs on is a second order logic machine.
The computer language is second order logic.
The assembler it compiles to is second order logic. [GOTO anyone?]
You can stick a first order algorithm on top if you like.
=======
 I’ve never seen a first order machine. I consider first order logic a nice useful fiction.
======
Yup. It is. It’s a practical fiction but it’s fiction.
=====
 It is, but machines have been built with second order logic based specifications.They do pretty good too.
======
 Specs are fictions until actualized by product. That’s all planning.
=====
Product is tangibles.
====
 Ah. That’s actually true. Its identity is a fiction.
=====
Fiction =/= bad or unwanted.
 ——-
 It can. In Church’s untyped lambda, TRUE=FALSE
=====
 Yup. That’s why it’s so powerful and we’re using it today.
======
You’re assuming consistency on my part.
======
 Identity is a fiction. So’s all representations. How it is.
======
Ultimately? It doesn’t. Ship of Theseus. So, it also does.
——
 Assertions form the basis for most truths if not all.
======
 You assert [x].
Is it true?
It’s true that you asserted [x].
Where you go from there depends on how you build your logic.
========
  Identity forms the base assumption of most philosophy, all math and most logics.
=====
 Time/change is what falsifies identity.That was the secret behind Church’s lambda calculus.Identity? It’s anonymous.
=====
 “At that moment”, yes. Including Time is what makes Church/Turing and all of computing work.Next clock tick? identity can be erased.
=====
 In computing it is. It’s all assertions. You assert a truth. So, now it’s true.
======
  You’re born, live and die. Cells constantly change. Parts and whole change continually. Brain states move while brain matter oozes around.At which point is your identity stable?At each change of thought, is that collective still you?Memory creates the fiction of identity.
——
 Assertion, really. But memory contains the assertion, just as writing down an assertion on paper creates identity.
======
 Einstein loathed quantum mechanics as it developed for it was based upon probabilities.Probabilities come from “ideal gas” which comes from games of chance.
======
 It is, although it uses a lot of handy tricks such as renormalization.Renormalization looks at ACTUAL measured values and sticks them into the equations, which recalibrates quantum theory to better fit reality.
=====
  Einstein was pure theory and verified by another.Quantum theories use real world measurements ALONG with theory along with probabilities and that mixture seems to work quite well.Kind of a self fulfilling prophecy.
——–
  Feynman thought it was cheating at first — it was him and two others who all independently came up with renormalization, each from a different angle and won them a combined Nobel Prize for it in 1962.http://www.riken.jp/en/research/inventions/renormalization/
——
 I’ve heard that Steven Weinberg reflects general consensus for how they think nowadays.” There is nothing arbitrary or ad hoc about the procedure; it is simply a matter of correctly identifying what we are actually measuring in laboratory measurements of the electron’s mass and charge.”– Steven Weinberg, Dreams of a Final Theory (1992), Chap. 5 : Tales of Theory and Experiment

=====
 That article from Japan highlighting Shinichiro Tomonaga, opened my eyes quite a bit. I always heard about Feynman – and he was a great, funny, “for the people” guy — but nothing about Shinichiro Tomonaga.
======
 What’s amazing to me is “super-many-time theory,” is IMPOSSIBLE to find online but for occasional references – one in 1946 extending it a bit… but it’s known….

..and YET, Western physicists keep reintroducing it as “brand new”…

https://arxiv.org/abs/0812.3869

—–

Leave a comment

Your email address will not be published. Required fields are marked *


× 3 = nine

Leave a Reply