Arthur (Skip) Gasch

Larry, Phil,

Regarding learning to organization information and anticipation, here's a point. If you walk briskly through a room, what you notice is what is moving, because that's how the eye organizes stuff. What's just hanging out the eye/brain dismisses. A guy in neutral clothing hang out by the wall and not moving (except glacially) is missed. Benevolent penetration of highly secure facilities by such people succeed.

Yet, if properly organized by a technology that builds a model based upon anomalies, causes such a person to stand out immediately. Humans haven't learned to do that well. Anticipation causes one to scan for what they expect, and minimizes their quality at discerning actual content. What do you expect to see in the picture below.

Did you see in it? If nothing, you're dead.

Jeff Daum

OK Skip, I'll bite: in the foreground, possible problems right and left of the path and an additional 3 or more as you follow the path out and to the right (2) and to the left (1) toward the back.

Philip Spiess

I won't answer, because I'm dead (even after referencing Richard Connell's "The Most Dangerous Game") -- but then, my eyesight isn't what it used to be (oh, wait! it never was! -- though never quite at the James Joyce level).

Jeff Daum

Paul, I think you are correct!  I read "in" as "if" and it clearly impacted my perception.

Paul Simons

Wow Jeff that was quick! And I'm heartened by your willingness to say that someone - anyone - with a different point of view, if supported by evidence, might be right. That's a welcome attribute in a post-truth, non-evidence-based milieu, which is I believe a French word and thus suspect in some quarters.

Jerry Ochs

I saw an elephant, a bunny rabbit, and a penguin.  I might be playing this game wrong.

Philip Spiess

But Jerry, where is the umbrella (or scissors)?  There is always an umbrella (either rolled up or open) in those pictures!

And Paul, you are correct!:  milieu is a French word; it means (in Middle English) "the toilet in the old mill" (cf.,  the phrase "Gardez l'eau!" interpolated into English as "loo").  The French couldn't understand why the English would put a toilet in an old mill, being used to using chamber pots and fireplaces (well, in the summertime -- don't ask) themselves, which is why they translate milieu as "environment" (probably with a question mark at the end).  The French themselves prefer, in such circumstances as, say, a toilet in an old mill, to use the phrase mise en scene, which means "a miserable scene" (vide a toilet in an old mill) or, sometimes, "surroundings of an event," which is a very polite and delicate -- and very French -- way of saying "going to the toilet in a mill."  [The English novelist George Eliot, who much preferred reading German to French, turned the whole thing on its head, putting a dent -- or  "dental" -- locus on the mill suroundings -- or "event" -- by focusing on one's mouth, rather than at the other end, when she wrote a book entitled The Mill on the Floss.  You might think that in this case it should read The Floss on the Mill, but she was doubtlessly translating from the German, in which everything is always backward, starting with the verb at the end of the sentence.  So geht es alles in eine fremde Stadt.]

Arthur (Skip) Gasch

Phil, Jeff, Jerry, and Paul,

A hint on the jungle trail pic. There are 3 threats, although if there had been others, they too would have been detected. Which side of the trail are they on? There is camouflage involved here.

I will ultimately show them, but have a question, "What AI algorithm could be used to detect threats, given that in one setting the "threat" may be very different than in another?" What if the setting was desert or arctic? What if the threat were different? In th pic I sent the threat is snippers.

Could one algorithm work for all threats in all settings? If so, what "attributes" would it have to have. It certainly could not be "rule-based." Does the content of the picture itself show us "where the needles in the haystack" are hidden? If so, how? If not "ruled-based" could it be a neural network? But if it was, would it have to be
"retrained" for each different "setting," it would have to work in? What would happen to the node-values of all the previous learning if something "new" was discovered?

The solution doesn't depend on increasing computational speed, or is information organization more important to intelligence. After all, CPU clock speed beats brain processing time, by a lot, but, the brain usually "sees" things that self-driving cars miss. What is that?

Could it be that no "algorithm" at all is required to see the "threat?"

It's an interesting discussion so far.

Art

 

Philip Spiess

Are we quite sure Skip Gasch's photograph wasn't taken in Krohn Conservatory?  (Just askin' -- in which case the threat would be the hazard of water on the slippery steps just after it opened in the morning.)

Larry Klein

Paul - boy, were YOU off-base. Those were Hong Kong hookers.  I never made it to Saigon during my tour over there.  Took my two R&Rs in Hong Kong and Sydney.

Skip - as to the pic, my eyes are failing me these days with retinal spots and cataracts, so the only threat I see is the likelihood that I'll trip on the rock in the path. 50 years ago would have been a whole different picture (to me, anyway). btw - I drove up old Paxton Ave last week, Not much is changed from our days there.

Arthur (Skip) Gasch

Let's take this back to the AI universe. How do you find a needle in a haystack, or in our case a snipper along the trail? By the differences of the properties. You need a method of storing data that catalogs differences. It is inherently "intelligent" and useful for all sorts of things.

Imagine a second picture of some white blood cells on a slide smear. If the information system has seen some cancerous ones, and has catalogued their property differences from non-cancerous ones, you can simply build a catalog of cancerous cell properties, and run the hundred thousand or so cells in a slide sample and instantly find the location of all the cancerous ones -- Making pathology tests a bit faster to perform

Does that make the approach AI or deep learning? It's analytics by example at least, and the
analytics queries are near real time.

Regards,

Art
 

Arthur (Skip) Gasch

Hey Paul,

about the length of rope, I didn't think it was Rudy, I was thinking more along the lines of Bill Clinton.
I think it will come to light whomever it was, seem to be something DOJ will pursue.
Should be interesting.

Art

Philip Spiess

Art:  Your most recent example has also brought the discussion back to my area of expertise -- deciphering/analyzing museum objects.  I constantly would tell my students "Look for the differences!  What distinguishes these two objects, which look basically the same, one from the other?"  (An exercise I gave them was this:  I took a large selection of screws, nails, bolts, nuts, wing-nuts, eye-screws, etc., of various sizes and kinds, into class and told the students to separate them into distinctive categories, as they saw fit; having been divided into a number of small groups to carry this out, the different groups often divided them in different ways -- which was perfectly harmonious with the activity.)

Having decided what makes them different, the next, and central, question was "why?"  Why did the maker (or the culture) decide to change something? did it make it work better? (practical change); did it make it look better? (aesthetic change); did it make it stronger, lighter, faster? (technological change); was it made out of cheaper or more available materials? (economic change) -- and so on.  Examining the differences (minute as they might be), and the likely reasons for them (not always knowable or deductible), was key to unlocking information about objects and their cultural relevance.

By the way, that picture on the left looks like a late 1950s map of the areas of semi-submerged land in south Florida that folks were trying to sell as subdivisions off ads on buses and early TV when we were in school.  (If you look closely, possibly under ultra-violet light, you will see the alligators -- or the umbrellas.)

Arthur (Skip) Gasch

Hey Phil,

Interesting.

Here's another wrinkle. Which items you find (in a dig for example) should be grouped with the other items, to form a 'set' of similar items.

That same issue arises in IT land, where a group of responses is plotted in any number of dimensions. The concept of "distance" is really about that, and gets very interesting as the number of dimensions increases. In two dimensions, it's as simple as getting a 'measuring' stick and using it.

In three dimensions, well that geometry or perhaps calculus, although the taught us calculus wrong at Walnut Hills, oh well.

What about in 7 dimensions or 100 dimensions? Every independent property of a thing is in fact a dimension. In the case of your nuts and bolts example, what the bolt was made of is a dimension. Was it wood, brass, iron, steel, lead... and what were the trace elements? If one digs through two archaeological layers, or if they intersect, and you have 12 bolts, that in one dimensional plain are some "distance" from each other, how do you group them into sets? What if they aren't at the same depth?

As TV reminds us, "Now we can drain the {ocean, land, whatever} and simply see them in 3-D (or in truth in N-D) so the issue of whether the lead bolt at depth 1, is part of the groups of 2 other lead bolts that are at depth 2 and 3, makes the exercise of grouping them.

Any thoughts about that? The notion of grouping items (in the museum) and how to detect the
forgeries from the real things? There is a simple way to do that.

Perhaps we should apply your technigues for identifying which of us humans could be differentiated into various property classes, based on subtle differences in how they present themselves. What do you think? Is there a sacred group and a secular group, and if so, what properties differentiate them? Are there degrees of secularness, and if so, does it matter?

Art

Philip Spiess

Art:

This is getting interesting.  The whole question of grouping objects is a complex one -- and one that one needs to be cautious about in museum work.  Early archaeologists in particular (such as Schliemann at Troy or Sir Arthur Evans at Knossos) were eager to group objects they found according to the preconceived notions they had of the sites they were investigating.  One hundred fifty or even one hundred years later, more recent scholarship and re-investigation of the sites often suggest that things were (or could have been) entirely "grouped" differently.  As museum professionals, we always spouted (and often adhered to) the concept of placing the object "in its context" -- but who decided what context or which context often determined how the object in question was interpreted to the public -- a remnant of Greek sculpture, for example, could be interpreted one way in an art museum (as an "artwork") and another way in a history museum (as a "signifier" of a period in an historical culture, such as ancient Greece, or as a token of an archaeological find).  A longtime director of the Corpus Christi Museum in Texas, Albert Heine, often gave his well-known speech on "Founding a Museum Based on a Nail," in which he would interpret a whole world of history, art, technology, science, and economics based, in turn, on the different contexts into which he could place the nail.

Now, although I am no "underground" archaeologist (though I was the founding president of the Washington, D. C., chapter of the national Society for Industrial Archeology, a largely "above-ground" archeological group), I have been associated with enough "real" archaeologists to know their professional methods.  When excavating an archaeological "dig" underground, they plot "distances" in three dimensions by establishing a horizontal grid of smallish squares on the surface, which they then extend downward through the soil in pre-determined measured levels.  It is of prime importance to them to note the "grouping" of objects within the squares, to determine the relevance of relationships between and among the objects ("artifacts") found.  Likewise, in most cases, each level further down in an archaeological dig is assumed to be a previous ("earlier") level of cultural activity, and thus the objects found there are likely to be dated to an earlier time, unless other evidences indicate otherwise.  (Heinrich Schliemann is supposed to have "discovered" -- or uncovered -- no less than seven layers of previous Troys.)  Needless to say, all of this rarely accounts for movements within the earth, including significant upheavals.  [I used to think that a useful exercise for training beginning archaeologists would be to have them sift through a cat's toilet box, eventually emptying it, but I never put the experiment into execution.]

And this is in some ways no different from the ways in which archivists operate.  Much of the time, when professional archivists are given or locate a significant cache of important papers, documents, stacks of bills, laundry lists, loose letters, interspersed journals, schedules of meetings, etc. -- i.e., totally unorganized, filed, or categorized documents -- they carefully move them in situ to where they will inventory them, leaving them in the stacked order in which they found them, because the "unorganized" order is a natural order of the way in which the papers' owner left them, and so there is often a connection among the papers located together.  Only after looking through those connections, and ascertaining them if possible, do archivists organize those documents by subject, date, or persons, or whatever system of organization they may be using.

Now, as to forgeries, that is a whole different subject!  (I'll try to be brief here.)  Forgeries of objects have existed since time immemorial, largely based on the presumed possible economic worth of said forgery.  (The Victoria & Albert Museum in London, a teaching museum of the decorative arts, purposely set up an exhibited collection of established forgeries in the 1850s, so that students could study them -- it's still there and is still used for those purposes.)  Of course, nowadays we have many scientific ways of testing objects thought to be forgeries, rather than just by sight -- but see some of the articles on the (New York) Metropolitan Museum of Art's famous "Etruscan Horse" sculpture -- after almost seventy years they're still debating:  Is it or isn't it a forgery, or at least not what it seems?  My own actual "real life" example was simply proved:  I had catalogued (first) the complete dental collections (going back to the 1830s) of what afterwards became the National Museum of Dentistry in Baltimore, and then catalogued (second) the complete dental collections of the National Museum of American History at the Smithsonian Institution.  (Cataloging museum objects is a complex process which I won't go into here, but computerization has eased much of the task.)  There was a reason I was contracted to catalog both collections:  the Baltimore museum's objects had been loaned for thirty years to the exhibits at the Smithsonian, and there was a serious dispute over whether all of Baltimore's objects had been returned by the Smithsonian.  All of the personnel involved in the initial loan agreement had died; I was hired to sort it out -- which objects belonged to which museum (all that I know about dentistry itself -- except that it hurts -- could fill a tooth).  The particular objects in contention were, historically, the earliest in the collection, and therefore (presumably) the most valuable:  they were two or three dental instruments of the 15th century, rare in any collection; Baltimore claimed them as theirs; the Smithsonian said they were theirs.  I studied them by sight at length for two or three days (see my several posts above about "Look and look again!" at the objects) -- they appeared way too new to me, but even antique objects can be "gussied up" to look fairly new.  Then it finally hit me:  the screws in the objects were an anachronism -- such screws did not exist in the 15th century!  These screws were too perfect; 15th-century screws were exceedingly clumsy and coarse by comparison.  So where did these instruments come from?  I had an inkling of a clue:  Baltimore (which at the time of its loan had no museum -- all its objects were in storage, hence its loan to the Smithsonian) had loaned its objects when the Smithsonian first set up its Dental Exhibit in the Old U. S. National Museum (now the Arts & Industries Building) before it was moved to the new (1964) National Museum of American History's "Hall of Medicine."  There had been an exhibits laboratory at the Smithsonian which produced replicas of objects which the Smithsonian needed to tell its historical story, but which it did not have in its collections.  Unfortunately, I found that the records of this lab no longer existed -- but the evidence was pretty clear to me:  these replicas (not forgeries, per se) had been manufactured by the Smithsonian to fill out its interpretive story.

As to differentiating human beings into distinctive or various classes, I am opposed to that:  despite the 39 "Ethnic Heads" that are sculpted on the outside of the original Library of Congress building (now called the Jefferson Building), chosen by the chief of the Smithsonian's Department of Anthropology (Otis Tufton Mason) in the late 1880s (and very interesting they are, too), I learned in my doctoral work on evolution, studying under one of the editors of the Darwin papers (located at Cambridge University), that the idea of "Human Race" does not exist in real science, but is merely a cultural construct -- and one that has not served humanity well.

Arthur (Skip) Gasch

Hi Phil,

Above are various instances of thee types of things, one red, one yellow, one blue.
The data has been fully normalized and is devoid of metric system scaling and affine distortions.
Now, how many "clusters" are there of each of the three colors? What "rule" determines how
to answer that question?

In some ways it isn't a fair question, because these things are higher dimensional, and have been
collapsed mapped onto a 2-D surface, but imagine they are locations in a 3-D space. Then again,
how many clusters or groups of yellow objects are there? and how many reds and how many blues.
Are there some singularities of yellow and red or blue ponts plus some clusters? If so, how many?

What is the frame of reference that caused the collapse mapping? Does that affect your answer?
Suppose the bottom right corner of the picture is the 'nearest' to you and the upper left corner is the farthest point away in this hypothetical 3-D space.

What if this is a 4 or n-dimensional space. How does the notion of nearest neighbor come into play? A cluster might be all objects of one color that have a distance to each other of less than K, a specific
distance. Would that make clustering easier and reduce it to a mathematical exercise?

Are the three yellow blocks singularities, or do they form their own cluster?
Red, blue and yellow here are multiple instances of only 3 properties. What if there were 1,000 colors because there were 1,000 parameters, or say 6 million colors in a genome (such as in the lifetime medical record the government is calling for)? It gets really interesting then?

Do you suppose people would be interested in investing in an AI solution for such problems, that could also expose relationships between the three colors and other outcome variables, such as the illness or wellness of the patient, or their survival at 2 years from a treatment, or their likelihood of readmission to a hospital ER within the next 30 days?

We have an information organizational structure that makes answering such questions simple algebra. We could be more specific under an NDA. Anyone interested in pursuing such matters. 

Art

Philip Spiess

Art:  Well, you've hit on the same set of inquiries, it seems to me, that I posed to my beginning museum students with my exercise of sorting out the bags of screws, nuts, bolts, nails, etc.  (To my knowledge, no one else in the entire set of professors of Museum Studies, past or present, in the whole United States -- including those in my own department at The George Washington University here in Washington, D. C. -- ever conceived of such an exercise, or thought that these were questions that even needed to be addressed.  Yet they open up a vast range of variable analysis and recognize that there are many things to be considered under the sun.)

You talk about "draining the ocean" to see certain things.  It was not until humans took to the air (and some time after that!) that archaeologists discovered that centuries-old underground remains of former civilizations (say, the Romans in Britain) could be easily seen or identified by the way in which grasses or other agricultural crops grew over them, even after millenia had passed, making them distinctly outlined areas when viewed from the air (no, I'm not talking about "crop circles").  Thus archaeologists knew where to dig.

On another note (we won't talk about "draining the Swamp" here in Washington, which is exposing all sorts of things), for centuries there had been stories and rumors about two large Roman ships, built by the Emperor Caligula (one absolutely palatial), that were sunk at the bottom of Lake Nemi, south of Rome.  Occasionally, over the many years in question, fishermen had brought up, say, a golden goblet or a portion of a statuette that suggested that something was indeed down there.  So finally, Il Duce, Benito Mussolini, in 1927 ordered Italian engineers to drain the entire lake in order to see what was down there.  And, lo and behold! there were, in fact, the two ships of Caligula that history had said were at the bottom of the lake!  The larger one included marble fixtures and mosaic floors, a heating mechanism, plumbing, and baths, technology that was not believed to have been possible until much later.  Mussolini had the two ships removed to Rome by 1929, and in 1936 the museum that was built to house them opened to the public.  However, World War II then broke forth, partly due to the machinations of Mussolini, and in 1944 the entire museum, ships and all, was destroyed by fire from Allied bombing.  Moral:  They existed for almost two thousand years, were recovered for fifteen, and then pffft!  O tempora!  O mores!

Philip Spiess

AND -- back to Skip's question:  How do you find a needle in a haystack?  Applying Occam's Razor to the question, and referencing Larry Klein's picture at Post # 4227, I proffer that, nine times out of ten, you find the needle by just casually settling back into the haystack (well in, mind you) -- and . . . the needle will find you!

Jerry Ochs

Sorry to change the subject but, does anybody remember this?

Philip Spiess

Jerry (you didn't quite change the subject:  you skipped from "Skip" to "Skippy"):  It does ring a bell, though I've never tried it.  (I do still use both products in my cooking -- but not together.)  What I do remember from the late 1950s is learning about "grilled peanut butter sandwiches" at Dick Ransohoff's house and having them there, and later having my mother, at my insistence, make them at home.  (I don't know whether this recipe was an invention of Dick's father, Jerry, who wrote a "Men's Cooking" column for The Cincinnati Enquirer, or not.)

Since Jerry has changed the subject, this may be a good time to consider other foods we enjoyed in our youth, which may be considered odd or weird by some of us, or considered now rather "out of fashion."  Yes, foods and recipes do go in and out of fashion -- witness the frozen foods and TV dinners of the 1950s, the nouvelle cuisine of the mid- to late 1970s, the petite plats of recent memory -- you'll notice how often these trends are French or "pseudo-French" -- and the current trend, at least in our major cities, of increasingly exotic ingredients, cooked (or not) in increasingly "daring" ways, and served to the diner in increasingly bizarre presentations (such shenanigans go back to the ancient Romans in the era of the writer Petronius and to the early Victorians in the era of the chef Careme).  We also have the current (and, in my reckoning, very good) trend of "farm-to-table" menus, and (in my reckoning) very pretentious and very expensive "chef's table" single sittings (with a lecture or demonstration by the celebrity chef in question) for the self-proclaimed "foodies" who attend.

So let us reflect on those "odd" foods (no, I'm not referring to those items left in the back of the refrigerator for several weeks, which suddenly seem to have turned purple and fuzzy), perhaps odd to us or others now, which we happily indulged in in our youth, but now unthought of, or no longer available, or just not eaten anymore.  [No discussions here, please, of Skyline chili or White Castle hamburgers or Graeter's ice cream; we've had plenty of those discussions, and such foods are not "odd," but are gifts from the gods.  (Possibly these were the "manna" from Heaven which the Israelites enjoyed after the Exodus from Egypt, although I don't think the chili or the hamburgers were made from quail, nor does Graeter's now make an ice cream that tastes like coriander seed that I'm aware of.)]  So I'll begin by listing some of the foods I used to eat, often at a Saturday lunch or Sunday dinner, foods which I haven't eaten in years.  (Somehow, a lot of these dishes are or now seem somewhat German to me, which should come as no surprise, given Cincinnati's and my family's ethnic heritage).

Beef Tongue Sandwiches (it's not easy to find beef tongue much any more).

Sardine Sandwiches on White Bread (the oil from the packed sardines sank so nicely into the soft white bread).

Calf's Brains with Scrambled Eggs (these were served to me at breakfast by Don Dahmann's mother after a number of sleepovers with Donald at his home in Vine Street Hill Cemetery -- and I ate them; they tasted something like sweetbreads).  Speaking of --

Creamed Sweetbreads on Toast ("sweetbreads" are the thymus glands and pancreas, usually of calves or lamb; my mother used to serve them on occasion and my father hated them, though I liked them -- I thought they were odd-tasting mushrooms; recently I've been able to order them as an appetizer in some of Washington's more exclusive restaurants).

Leber Klosse Zuppe (soup with liver dumplings; my great-grandmother, and later my mother, used to make this; you also could order it at the Temple Delicatessen, across from Shillito's).

Schmierkase ("smeary cheese," a very soft, small curd cottage cheese with a fair amount of whey, slightly sour-tasting, which we used to get from Coors' Dairy on Gray Road in Winton Place; you could also order it at Grammer's German Restaurant downtown).

Saurbraten ("sour roast," a German specialty made by marinating roast beef in a sweet-and-sour sauce, often flavored with ginger snaps, for several days, then simmering it in the sauce for several hours; my grandmother had the best; I have her recipe -- no German restaurant I've eaten in has ever come close, except maybe The Student Prince Cafe in Springfield, Massachusetts, or Schatzi's in Baltimore, now closed; my grandmother used to serve it with red cabbage and potato pancakes with either the Saurbraten gravy or Amish apple butter over the pancakes -- my favorite meal when I came home from college!).

Goose (I haven't had goose in years; my great-grandmother used to serve it at Thanksgiving).  Speaking of which --

Molded Ice Cream (these would be the desserts at my great-grandmother's Thanksgiving dinners; they were molded in the shape of turkeys and had colored, flavored ices imbedded in the ice cream; she didn't make them, but I don't know where she bought them; I've occasionally seen metal ice cream molds in antique stores, but I've never seen molded ice cream for sale ever again).

AND FINALLY (I'll stop here):  Schwartenmagen ("headcheese," i.e., meat, fat, and spices in aspic -- which my family pronounced in true "Corryville Dutch" fashion, "Schvattamagga" -- they also pronounced "wurst" "woosht"; my grandmother loved it, and I, very occasionally, would join her in eating it on crackers).

Jerry Ochs

From Wikipedia:

Goetta is a meat-and-grain sausage or mush^[1] of German inspiration that is popular in the greater Cincinnati area. It is primarily composed of ground meat (pork, or pork and beef), pin-head oats and spices.^[2] It was originally a dish meant to stretch out servings of meat over several meals to conserve money,^[3] and is a similar dish to scrapple and livermush, both also developed by German immigrants.^[4]

Limburger (in southern Dutch contexts Rommedoe, and in Belgium Herve cheese) is a cheese that originated in the Herve area of the historical Duchy of Limburg, which had its capital in Limbourg-sur-Vesdre, now in the French-speaking Belgian province of Liège. The cheese is especially known for its strong smell caused by the bacterium Brevibacterium linens.^[3]

Paul Simons

Just a quick addition - here in the Philly area Goetta is known as Scrapple and revered as a breakfast food. It must be around New York as well - there's a Charlie Parker tune called "Scrapple from the Apple" - that particular apple being  NYC, "The Big Apple", the thing that the biblical serpent used to get Adam and Eve to get into sinning, which evidently NYC has taken to skyscraper levels, But for Charlie Parker it was about pure joy, it had nothing to do with anything else. 

Jerry Ochs

When our two sons were in Cincinnati for the first time, my family introduced them to the Brown Cow (vanilla ice cream and Barq's root beer).  Did anybody else enjoy this simple treat?

Philip Spiess

Yes; we usually made them at home.  Curiously enough, about five years ago, when Browne Academy sent me as one of the chaperones on the 8th Grade trip up the Hudson River, we ended up for lunch in a massive, elaborate classic diner (with all of the neon, mirrors, and jukeboxes appertaining thereto) at Hyde Park, across the road from FDR's home.  There on the menu was a "Brown Cow" (though I don't think it was made with Barq's); I immediately ordered one, and when it came, it was huge (but good!).

Chuck Cole

I remember brown cows being vanilla ice cream and coca-cola.  Black cows were made with root bear and pink cows with red cream soda.