Author: Maverisk

Yes, hopefully, you were thinking of some Great music; here. Otherwise, I wanted to ask, again, whether anyone would have pointers to some easy intro on modelling – on how in neural networks, the weights are the model. In terms of parametrising a SAP system, but learning by training and doing; changing the parametrisation as you go in production, to accountants’ delight [maybe not so much]. This, compared to Expert Systems and evolutionary algo’s. The latter, so much underestimated but hey, let’s get something going in that area and first try to get some catch-all explanation-for-the-masses on neural networks out, right?

Then again:

[Emerging-from-software library; Dublin]

“Asking for a friend” whether anyone would have a lead on how things stand with AI doing bug hunting ..?

As [hey, by sheer coincidence, exactly] three years ago I asked already. And got no serious response then. And since, there have been more talk about AI on either side of the infosec FLOT, but not much on How. Or where the FLOT is. Throw in some Adversarial Examples and there you go; arms’ races and falling behind in those…

By now, it should be easy to do such a thing; have an ML system learn what sloppy coding looks like, even when that is hiding in plain sight i.e., OK on first (syntax/compile) impressions but functionally tricky ..? A (data/input) edge case quality break-down severity tester would also help.
Then again, I suspect that some masters of invention haven’t mastered the solution, yet… As per DARPA’s request for your proposal to help solve a tangent issue; had they solved the above, they would’ve been able to translate the results to the latter.

In return, for your viewing pleasure:

[Just a pretty picture – when you’re an engineering-inclined mind; Binckhorst Den Haag Voorburg..?]

Owww that is great! Banks’ bank(let) going out to tell everyone how great their new regulation is! [I know, it’s not theirs; they’d dream it were]

Maybe ‘great’ isn’t the word. But a fact is that if anyone involved in banking advises you anything, you do the exact opposite – their only interest is their own which, given the bilateral relation in any trade, means you’re getting robbed blind if you listen.

They even tell you that giving more parties access to your information, improves the security over it. Yes they do, that’s no lie. It’s unlike hips, this page, 4th bullet.

No wonder the earth’s population minus the handful of ‘bankers’ [apologies for the atrocious word: ] pegs bankers at the bottom of any ranking of social status, admittedly not far below pole dancers at a same-sex SM swingers’ club.

On a positive note:

[NO THE CLIENT IS AT THE TOP; Girona]

Great music. Dystopian future:

pic.twitter.com/0NlKgJCSPV

— Lost Weapons (@LostWeapons) April 3, 2019

Either those drone’lets are weaponised or the blimp is monopolised. No escape. Unless … this.

Plus:

[Adversarial but irrational examples, maybe that’ll do the trick; Baltimore]

Quite some time ago, I posted a thingy on steganography and how the Internet is not filled with ..x pics but cat pics, for that reason.
More recently, I posted a bit on how traditional crypto and stego (in the form of mimicry and crypsis, see here) are comparably used in adversarial-example AI examples. Not discounting the wider definition of the term [link will work post-June19 ;-/ see here if you’re overly eager], but here, in a narrower sense.

Again, AI may come to the rescue at picking up traditional stego in any format, if somewhat-broadly or -narrow pointed to that qua learning.
It may also be trained to detect any tampering towards adversarial (narrow sense) so such tampering may become less useful ..? Possibly, plausible deniability [oh how great a buzzword to have all but lost its popularity] may come into play; but hey, when the AI overlord system says so, who are you human to derogate it ..!?

Somehow, I feel we aren’t yet on a sufficiently wholesale track with this – though humans cannot detect pixel-wise tampering, they can detect it on a much higher abstraction layer than AI can / will(??). Since evolution has favoured the blurry-yet-Kant’ian sense-making on sum-total (visual [1]) impressions, not bit-wise adversarial anomaly detection; for reasons of economy and speed, one guesses – we’ll have to ask evolution when we meet it which given the speed of AI developments may be quite soon…
Au contraire, AI ‘gets’ pixels but not yet too much of full picture decomposition into ‘real’ objects. The things that Kant saw, but e.g., Buddhism denies. [2]

Now, I’ll leave you to ponder it all, with:

[Explain that!]

[1] Surprisingly, Kant doesn’t go into other senses all that much, and uses the visual to ‘depict’ [recursive pun accidental but also intended] impressions as if 24/sec. slides.
[2] Both have a problem with Time, though. Kant has a ‘a priori synthetic’ categorisation but makes it fundamental for Vernunft to exist at all (as time is required to make sense of, i.e. decompose, impressions) and still doesn’t define time properly, and assuming such a thing as external reality / ‘objects’ as ‘real’ things out there but ‘space’ not existing. Buddhism does without space or external reality altogether, but has inherently cyclical time and an escape from that ..? Possibly I’m conflating various sub-strands here.

Please good sirs and ladies, may we get back to serious business and drop the exitespeak of Robotic Process Automation (here), as it’s all very nice you know an abbreviation but when there’s no substance behind it, drop it.

Why dweeb with ‘robots’ in an area where there’s no such thing?
The original demonstrate-once, repeat-indefinitely kind of ‘training’ (one size fits all, mostly) had to do with robots indeed, e.g., in factories where robots have been around for decades now and such guided ‘programming’ (hardly) was easy. Yes, that’s right, actual robots were around. For decades already so referring to that makes you an extreme laggard extraordinaire.
Now, however, the RPA thing is applied to pure algorithmic, doing-away-with-physical stuff. Not robots, eww! they’re things with grease and we only want to deal with nice GUIs. But still we want the ring of being Interesting with our jargon.

Simpletons. Just don’t.
Call it digitisation and you’re done. Or workflow pattern extraction or whatever. But leave the robots to the people that actually have an impact on the world.
At the very least, study this. And realise that the above, is in there – a lot [here].

OK?

And, unrelated to anything, this.

[Designed by not robots you … (censored), just humans with trivial tools – but then, designed: appropriately; Toronto]

Despite apparent suppression of the signals for purposes of bread and circuses, there is a thing called Downturn looming.
Which, rightfully noted, calls for prep. Of the serious kind. Yes, one needs to get an umbrella when it doesn’t rain yet, or they’ll be out of stock. Or hugely expensive. Or be found loaned from a banker.

Not only train [motto also of 323sqn but huh, and this doesn’t quite match as could somehow be expected …!]], which is about getting used to dealing with the difficult and unexpected so that no matter what’s thrown at you, you know you’ll manage. That’s about resilience, with another leg in having prepared evasive manoeuvres though mostly about having stashes of money to be able to adapt when that’s called for.
But also ensure robustness in the first place, corrective manoeuvres that teflonise your organisation when something hits, though mostly about having stashes of money to be able to weather the storm when that’s called for.

In general, read the above-linked HBR piece and this book – almost forgotten it seems when now, it becomes relevant again.
And then, act. Be prepared. Long-termism is about to win again.

For now, …:

[Before the French have taken it; … sorry, this being Monteriggioni or hey am I sure?]

Aha! This Friday’s Thought doesn’t give you as much of a problem, as it gives a Revelation:
That’s why the others moved ahead of you …

As is displayed in, yet again, a Farnham Street’s article. That you should read, and then see how you fit in, you misfit </endearing>, as you moved through life.

And now, for something completely the same; a picture…

[Segovia castle; now I’ve done. Sorry, am done. Not theoretically perfect, but it’s here, this post. With this (own, unedited) picture]

Why is this Dutch piece even news, still …? Physical paper ‘data’ leakage, of highly sensitive information intell dare we say. Edited to add, this one (also unreadable to most).

a. No we may not; b/c that is reserved for serious cases not trivia [on a non-trivial scale]
b. Yes we may, but still, why is this newsworthy?
c. Because it isn’t. It’s just flooding. Edited to add, the problem is Huge, as per this Bruce Schneier insight i.e. Truth.

The c. part is the tricky one. On the one hand, dismissing the ‘news’ as being nothing newsworthy, superfluous reporting of trivia. On the other, apparently still necessary.
Necessary that is, to point out that mistakes will always be made. Keep things fresh, even. Otherwise, 1. slacking off will happen, 2. progress will never be made.

I’ll stop now, to not add to the blurb. With:

[Sleep tight; Torun]

Earlier, I mentioned the ‘latest’ developments in physics, being the denial of such a thing as an objective reality ‘out there’. Only your thoughts remain, though they include the whole Universe. This was already summarised in this after the foundation in this, but mostly forgotten as the simple summary was an ouroboros; the original not so much but not fathomable by some simpleton minds, maybe.

Now, new evidence has surfaced – where does it exist? – that actually, that’s right: There is no such thing as objectivity. Thus destroying all that the new physics had achieved so far. Because the end conclusions backtrack to some very deep foundations, that hence also have to be undone, unwound.

Which would be academic, were it not that e.g., quantum crypto based on entanglement based on quantum physics, may also have to be undone as part of the Gödelian ‘nihil sequitur’. If ever there was a summary of the impact of Gödel, there it is in the latter short form.
Same, possibly, with quantum computing. The blizzardly faster it seems to be, the more theoretical problems arise with programming, running, and reading the results of computations with it. Theoretical not in a sense [Sense? What’s that ..?] of ‘only occurring in theory but practically irrelevant’, but of being able to be proven from theory to occur everywhere in practice.

Possibly, the Man will still be proven to be right …

Ah. All semantics. Indeed…

In unrelated news:

[What is space-time anyway? Noordwijk]