Some "peekabo" shots from early morning Jully 1, 2021

Geometric foundations of Deep Learning

Geometric foundations of Deep Learning https://t.co/B97gUekU09 #DeepLearning

— Machine Learning (@machinelearnflx) August 22, 2021

Nam June Paik, King of all haiku

"I am the greatest Haiku poet of all periods since 1967, and it is scientifically provable."

Nam June Paik, 1968 pic.twitter.com/qlZ1Ynbftq

— James Ryan (@xfoml) August 20, 2021

OpenAI's Codex demo [machine learning in its natural environment]

Interesting. A major weakness of these various machine learning systems is that they have, at best, limited access to the real world, and so they cannot learn in the context in which humans ordinarily learn. We're born primed to deal with that world with millions of years of evolutionary history behind us. Machines, alas, are a blank slate. The obvious place to apply machine learning, then, would be the world of computation. That, after all, is the NATURAL world for computers, their native environment. That, apparently, is what Codex has done.

As I noted last year, when giving my initial quick take on GPT-3:

Think AI as platform, not feature (Andreessen). Obvious implication, the basic computer will be an AI-as-platform. Every human will get their own as an very young child. They're grow with it; it'll grow with them. The child will care for it as with a pet. Hence we have ethical obligations to them. As the child grows, so does the pet – the pet will likely have to migrate to other physical platforms from time to time. [...]

So, the AGI of the future, let's call it GPT-42, will be looking in two directions, toward the world of computers and toward the human world. It will be learning in both, but in different styles and to different ends. In its interaction with other artificial computational entities GPT-42 is in its native milieu. In its interaction with us, well, we'll necessarily be in the driver's seat.

Where are we with respect to the hockey stick growth curve? For the last 3/4 quarters of a century, since the end of WWII, we've been moving horizontally, along a plateau, developing tech. GPT-3 is one signal that we've reached the toe of the next curve. But to move up the curve, as I've said, we have to rethink the whole shebang.

We're IN the Singularity. Here be dragons.

My first thoughts on AI-as-platform, though I didn't use that term, can be found in a working paper containing ideas dating back to January of 2006, PowerPoint Assistant: Augmenting End-User Software through Natural Language Interaction, https://www.academia.edu/14329022/PowerPoint_Assistant_Augmenting_End_User_Software_through_Natural_Language_Interaction. Here's the abstract:

This document sketches a natural language interface for end user software, such as PowerPoint. Such programs are basically worlds that exist entirely within a computer. Thus the interface is dealing with a world constructed with a finite number of primitive elements. You hand-code a basic language capability into the system, then give it the ability to ‘learn’ from its interactions with the user, and you have your basic PPA.

Friday Fotos: Double vision at 51 Pacific

Rhythm and the perception of syntax

In short, with @MicahGoldwater, we tested whether the alignment of metrical stress patterns (independent from regularity) could influence how robustly people comprehend syntactically complex sentences. pic.twitter.com/Z4uXvAJd2w

— Courtney Hilton (@courtneybhilton) August 19, 2021

Experiment 2 then replicated this for listening to auditory speech!

(yellow == congruent; orange & red == incongruent) pic.twitter.com/pcMfJ1csbJ

— Courtney Hilton (@courtneybhilton) August 19, 2021

I think these results will be relevant for work making sense of neural oscillations in language processing, especially low-frequency delta oscillations and their relation to syntactic/compositional structure and memory (@davidpoeppel @andrea_e_martin @johannarimmele )

— Courtney Hilton (@courtneybhilton) August 19, 2021

+ for making sense of why when silently reading language, imagining the text in a vivid prosodically rich way aids comprehension (@GolfWriterKiel), and how imagined metrical prosody can affect syntactic parsing (@KentnerGerrit).

— Courtney Hilton (@courtneybhilton) August 19, 2021

Anyway, lots more discussion in the paper if you're interested and I'm more than happy to answer any questions!

— Courtney Hilton (@courtneybhilton) August 19, 2021

Babe of the future [51 Pacific]

Reading Spacecraft 4: The future

I wonder if he’ll say something about the future? I thought when I asked Tim for a review copy of Spacecraft. The future sort of goes along with science fiction, no? But Tim is also very much interested in climate change and the Anthropocene, and they entail the future too.

And in a special way. For climate change is what Morton calls a hyperobject, an object so extended in time and space that we cannot grasp it (easily if at all). If we think of climate change as one single thing a hyperobject, why then an undetermined fraction of it hasn’t even happened yet, most likely a larger fraction than we’ve so far experienced. Yet if it is all one (hyper)object, one thing, then there is a sense in which THAT future is now. Is the future but a dimension along which objects (can) withdraw?

If Tim’s going to talk about spacecraft he just has to talk about the future. And he does (p. 73):

The future is everywhere, you can reach it any time–the real future or as I like to say sometimes, the future future. Futurality is the possibility that things could be different. [...] The hyperspace of Star Wars is the closest a regular cinema-goer gets to realizing that film isn’t just about the visible, but about the invisible, and in particular, the feel of motion, all the liquid celluloid spooling through the projector [...]. It’s a lovely gift that George Lucas gave to ordinary people–something utopian to keep them going.

Yes.

Tomorrow, next week, three months from now, 13 years from now those times have not happened yet. But that by itself does not situate them in the future. Futurality implies a certain kind of relationship to the present. [Does this mean the Nietzsche’s idea of the eternal recurrence is a denial of futurality?]

I put the issue to some of my humanist friends on Twitter and they replied that the future, “as something that can and will be significantly changed by human effort” (my formulation) is the product of the late 18th century. The future is a kind of place that exists at some time other than the present, some time after the present. Maybe it’s some kind of fold in hyperspace.

Now consider this passage from Morton’s last chapter (“Anyone” p. 101):

Stealing the Falcon and playing it like an instrument is a nonverbal image of revolution, like those in Sergei Eisenstein’s film about the Russian Revolution, October.

But unlike in the Russian Revolution, it’s not about some artist-like hero seizing the moment (Lenin). It’s about people from the criminal class whom Marx sometimes calls the enemy! And they’re seizing an instrument and playing it, listening to it, not imposing their will. America is a country that isn’t really one because of slavery. America is (from) the future. If it’s something from the past, it can only be a hotbed of fascism. If America is a weird, junky model of future worlds in which there is planet-scale human collective awareness and action, then ... well the Falcon is the utopian quality of America itself.

It’s this sentence that caught my eye: “America is (from) the future.” Read it both ways, with and without from, at one and the same time. Without it’s saying the future is going to be like America is now. With it’s saying America will be birthed at some date (way) ahead of the present and is heading for us (at warp 9).

But what is America? It’s probably another one of those hyperobjects. So who knows what or where or how it is. But somehow it’s become entangled with the Millennium Falcon.

A final thought: The original Star Wars opens “A long time ago in a galaxy far, far away...” When and where is that? What happened to the future? Is it lost?

Other posts in this series: https://new-savanna.blogspot.com/search/label/Morton-Spacecraft.

XKCD on machine learning

 

Increasing harmonic richness in Western Classical music

"we find that [harmonic] vocabulary richness undergoes a clear increasing trend across the history of music, as expected from qualitative musical wisdom. "https://t.co/etIEZJDCMo pic.twitter.com/XxiUMeo7Ad

— Alberto Acerbi (@acerbialberto) August 19, 2021

The abstract for the linked article:

Music is a fundamental human construct, and harmony provides the building blocks of musical language. Using the Kunstderfuge corpus of classical music, we analyze the historical evolution of the richness of harmonic vocabulary of 76 classical composers, covering almost 6 centuries. Such corpus comprises about 9500 pieces, resulting in more than 5 million tokens of music codewords. The fulfilment of Heaps’ law for the relation between the size of the harmonic vocabulary of a composer (in codeword types) and the total length of his works (in codeword tokens), with an exponent around 0.35, allows us to define a relative measure of vocabulary richness that has a transparent interpretation. When coupled with the considered corpus, this measure allows us to quantify harmony richness across centuries, unveiling a clear increasing linear trend. In this way, we are able to rank the composers in terms of richness of vocabulary, in the same way as for other related metrics, such as entropy. We find that the latter is particularly highly correlated with our measure of richness. Our approach is not specific for music and can be applied to other systems built by tokens of different types, as for instance natural language.

See my Stages in the Evolution of Music, Journal of Social and Evolutionary Systems, 16(3): 273-296, 1993, https://www.academia.edu/8583092/Stages_in_the_Evolution_of_Music.

Transcendent experience in performing jazz [Mulgrew Miller]

I've been collecting accounts of such experiences. That document has now reached its 6th version. I suppose that one day I'll transcribe some of Miller's words and append them the the document.

Emotion and Magic in Musical Performance, Version 6, https://www.academia.edu/16881645/Emotion_and_Magic_in_Musical_Performance_Version_6

H/t Bill Berry.

Out back at 51 Pacific

Martin Luther on music: "Music drives away the devil and makes people happy"

Forward to Georg Rhau's Symphoniae, a collection of chorale motets published in 1538, as follows:

I, Doctor Martin Luther, wish all lovers of the unshackled art of music grace and peace from God the Father and from our Lord Jesus Christ! I truly desire that all Christians would love and regard as worthy the lovely gift of music, which is a precious, worthy, and costly treasure given to mankind by God. The riches of music are so excellent and so precious that words fail me whenever I attempt to discuss and describe them.... In summa, next to the Word of God, the noble art of music is the greatest treasure in the world. It controls our thoughts, minds, hearts, and spirits... Our dear fathers and prophets did not desire without reason that music be always used in the churches. Hence, we have so many songs and psalms. This precious gift has been given to man alone that he might thereby remind himself that God has created man for the express purpose of praising and extolling God. However, when man's natural musical ability is whetted and polished to the extent that it becomes an art, then do we note with great surprise the great and perfect wisdom of God in music, which is, after all, His product and His gift; we marvel when we hear music in which one voice sings a simple melody, while three, four, or five other voices play and trip lustily around the voice that sings its simple melody and adorn this simple melody wonderfully with artistic musical effects, thus reminding us of a heavenly dance, where all meet in a spirit of friendliness, caress and embrace. A person who gives this some thought and yet does not regard music as a marvelous creation of God, must be a clodhopper indeed and does not deserve to be called a human being; he should be permitted to hear nothing but the braying of asses and the grunting of hogs.

and

.... if one sings diligently with skill and application, then music can make man good and at peace with himself and his fellows by providing him a view of beauty. Music drives away the devil and makes people happy; it induces one to forget all wrath, unchastity, arrogance, and other vices, quia pacis tempore regnat musica (for music reigns in times of peace).

The US didn't know what it was doing in Afghanistan [It doesn't know what it's doing in the world]

Jeffrey D. Sachs, "Blood in the Sand," Project Syndicate, August 17, 2021.

The article opens:

NEW YORK – The magnitude of the United States’ failure in Afghanistan is breathtaking. It is not a failure of Democrats or Republicans, but an abiding failure of American political culture, reflected in US policymakers’ lack of interest in understanding different societies. And it is all too typical.

Almost every modern US military intervention in the developing world has come to rot. It’s hard to think of an exception since the Korean War. In the 1960s and first half of the 1970s, the US fought in Indochina – Vietnam, Laos, and Cambodia – eventually withdrawing in defeat after a decade of grotesque carnage. President Lyndon B. Johnson, a Democrat, and his successor, the Republican Richard Nixon, share the blame.

In roughly the same years, the US installed dictators throughout Latin America and parts of Africa, with disastrous consequences that lasted decades. Think of the Mobutu dictatorship in the Democratic Republic of Congo after the CIA-backed assassination of Patrice Lumumba in early 1961, or of General Augusto Pinochet’s murderous military junta in Chile after the US-backed overthrow of Salvador Allende in 1973.

In the 1980s, the US under Ronald Reagan ravaged Central America in proxy wars to forestall or topple leftist governments. The region still has not healed.

And so on.

What these cases have in common is not just policy failure. Underlying all of them is the US foreign-policy establishment’s belief that the solution to every political challenge is military intervention or CIA-backed destabilization.

That belief speaks to the US foreign-policy elite’s utter disregard of other countries’ desire to escape grinding poverty. Most US military and CIA interventions have occurred in countries that are struggling to overcome severe economic deprivation. Yet instead of alleviating suffering and winning public support, the US typically blows up the small amount of infrastructure the country possesses, while causing the educated professionals to flee for their lives.

Even a cursory look at America’s spending in Afghanistan reveals the stupidity of its policy there. According to a recent report by the Special Inspector General for Afghanistan Reconstruction, the US invested roughly $946 billion between 2001 and 2021. Yet almost $1 trillion in outlays won the US few hearts and minds. [...]

In short, less than 2% of the US spending on Afghanistan, and probably far less than 2%, reached the Afghan people in the form of basic infrastructure or poverty-reducing services. The US could have invested in clean water and sanitation, school buildings, clinics, digital connectivity, agricultural equipment and extension, nutrition programs, and many other programs to lift the country from economic deprivation. [...]

The US should never have intervened militarily in Afghanistan – not in 1979, nor in 2001, and not for the 20 years since. But once there, the US could and should have fostered a more stable and prosperous Afghanistan by investing in maternal health, schools, safe water, nutrition, and the like. Such humane investments – especially financed together with other countries through institutions such as the Asian Development Bank – would have helped to end the bloodshed in Afghanistan, and in other impoverished regions, forestalling future wars.

I fear foreign policy is more driven by projective needs than by rational calculation of any kind.

Decoding neural data (from real brains) by using deep learning

1/7: Although DL methods are highly promising for cognitive decoding, their widespread use is generally hindered by:
1) their lack of interpretability,
2) difficulties in applying them to small datasets, and
3) difficulties in ensuring their reproducibility and robustness.

— Armin Thomas (@rmin_thomas) August 17, 2021

3/7: We argue that the explanations of backward decomposition approaches, such as layer-wise relevance propagation and DeepLIFT, are best-suited to faithfully reveal the features of the neuroimaging data that underlie the cognitive decoding decisions of DL models.

— Armin Thomas (@rmin_thomas) August 17, 2021

5/7: We argue that weakly and unsupervised learning techniques, such as data programming or contrastive learning, are well-suited to pre-train DL models at scale across vast and diverse public neuroimaging data: pic.twitter.com/x9KeQHYBCg

— Armin Thomas (@rmin_thomas) August 17, 2021

7/7: To improve the reproducibility of DL model performances and to make them more robust towards the diversity of real-world data, we make specific recommendations, for example, to randomize as many aspects of the training pipeline as possible when evaluating model performances.

— Armin Thomas (@rmin_thomas) August 17, 2021