Complex Geometry in Architecture
Sponsored images from 
View allAll Photos Tagged Computational
Playing around with the built-in computational ND filter in the OM-1, in Donard forest, Newcastle. Turns out it’s pretty good…
Computational portrait photography of this flower at Seattle Center—site of the 1962 World’s Fair.
Did you know you can dial in the aperture after the fact? Me neither. Next time I’ll layer tight aperture over wide, to get the full blossom like this, and a super-smooth background.
long stories shortened... (discarded and abandoned and intertwined short stories) well..actually they are chunks and fragmets and notes of stories that never made it
____________________________________________
a young PhD math candidate writing his dissertation on an obscure arab mathematician from the middle ages who specialized in cycles and periods in infinite series and develops a process to determine prime number density in a large number space. (which is all and good) except this makes it an excellent tool to decrypting military grade encryption, which is based on the computational difficulty of factoring large numbers into their prime components
the arab mathematician was ultimately censured by the religious mullahs for developing tools to rationalize the infinite, which is of course the nature of Allah and for man to attempt to place Allah into a human scale is blasphemy
so the arab mathematician disappears and the young phd candidate finds that his dissertation has been suspended pending review but cant get any information on who is reviewing it
finally another young mathematician approaches him and starts a long discussion on math and the nature of numbers and the mathematicians love of the underlying structure of reality that math represents. the phd candidate is leary of this mathematician cause he wont answer what he does or where he went to school or how he knows so many cutting edge fields in math
eventually, the young mathematician offers the phd candidate a position with the NSA, National Security Agency, (where all the big crypto and high math goes on) but explains that if he accepts that he will essentially disappear from his current world. his work will be classified, he will not be able to publish in academic journals or speak in public, or talk about his work to his friends on the outside, but the compensation is that he
would be able to work unfettered with the greatest math minds in the country, totally funded, free to explore any field or fancy he thought. after a few moments of thought, the phd accepts.
then the story will go back to the arab mathematician who is also approached my a young beared mullah, who offers him a position within his group of thinkers who do ponder and explore the nature of nature reality and Allah through mathematics, but that by joining them he would need to disappear from the world, after a few minutes of thought, he too accepts...
--
Daniel sipped his 6th coffee (colloidal suspension for caffeine transport) while his batch jobs on ramanet, the Indian supergrid, finished their checksum verification. His chin, a bit stubbly, itched. His eyes, a bit red, were sore. The goa trance shoutcast feed had mushed into a fast cadence drone. The flat screen monitor warped and bulged with the oscillating fan blowing on Daniel's face
'O' glamorous larval life of a PhD student...' he jotted and doodle-circled on his notepad.
Daniel cracked his neck and jutted his jaw, stretching out the accumulation of kinks, as RamaNet finished the final integrity check on his dataset. this two hour round of processing on the Indian supergrid would cost about $130 out of his precious grant fund, but you couldnt beat the bargain. 120 minutes times 150,000 PCs in the RamaNet processing collective = 1,080,000,000 seconds or 18,000,000 minutes or 300,000 hours or 12500 days or 34.25 years of processing time for the price of a video game. Calculation was commoditized now. You uploaded your pre-fromatted dataset to RamaNet. the data was packeted and sent to out to 150,000 Indians who lent a few percents of never-to-be missed CPU cycles off their systems for background processing. when their alotted package was completed it was sent back to RamaNet for re-assembly into something coherent for the buyer. in return the Indians got a rebate on their net access charges or access to premier bollywood galleries or credit towards their own processing charges. a good deal all the way around. Daniel's dataset, an anthology of complex proofs from a long-dead arab mathematician, was queued with amateur weather forecast modeling, home-brewed digital CGI for indie movies, chaos theory-based currency trading algorithms, etc. the really high end, confidential jobs, like protein folding analysis or big pharm drug trials were more likely handled by the huge western collectives of several million collaborative systems, usually high-performance machines in dedicated corporate server farms. the cost there was out of Daniel's range, but you got a faster return and better promises of encryption for your buck.
Daniel scratched his scalp and flexed his fingers. 'two months from today i will be a doctor of mathematics...and no job. damnit. i need to find something fast.' Daniel calculated in his mind how quickly the student loans repayments would kick in and completely wipe him out. RamaNet would have done it in nanoseconds, ha! he laughed to himself. Daniel had avoided the rounds of job interviews and recommendations that passed his way. he was too absorbed in his research to look ahead, and perhaps a bit intimidated by the idea of the job hunt flea market. flexing his CV, getting a monkey suit, trying to explain his research to recruiters, who were often the same finger-counting business majors in college that made his skin crawl. Daniel always felt a bit embarrassed when he announced he was math PhD candidate. folks would immediately glaze over,
tsk tsk out a 'that's interesting', and swiftly change the subject. something will come up, he mantra'd to himself over and over, something will come up. stick with ali, there is something real in there, just a bit deeper. the real problem was his thesis advisor. dr. fuentes was not returning his calls, his secretary was not taking appointments from Daniel. he had submitted his finished draft of his thesis two weeks ago, but hadnt heard back since, except for a cryptic email saying that the review committee was having some issues with his paper and that Daniel would be hearing from him shortly. Daniel was rerunning his calculations on RamaNet to assuage the gnawing doubt that he completely botched some component of his argument and that the review committee was debating some manner of telling him to redo the entire effort. no PhD and no job. that would ice the cake. Daniel started calculating his body mass and general aerodynamic resistance relative to the height of the school cathedral to figure out if he had time to reach a terminal velocity before impact...only a failed math PhD would attempt to determine at what speed his body would smack concrete, he morbidly thought to himself.
ali ja'far muhammed ibn abdullah al-farisi slipped meditatively on his cup of water, thinking about his proof. he dipped a finger in the cup and held up a droplet of water under his fingertip, watching the sunlight prisimatically splay out on the mouth of the cup. 'praise be Allah and his wonderous bounty' he mumured to himself.
the elders had been in conference all day over his proof. though the heavy doors to their chamber were closed, he would occasionally hear muffled but distinctly angry shouts. ali sat on a divan in the anteroom, served numerous cups of tea by an obviously nervous secretary. ali knew there was deep resistance to his research, but for the life of him he couldnt figure out why. he was a simple mathematician. he came up with some unique observations. he wanted to share them with his peers...
_____________________________________________________
Overview: biotech researcher discovers a new life-extension technology and is murdered. He is cryogenically frozen for 150 years. When he is
revived he must stop a dark corporate conspiracy – and find his murderer.
Summer 2015 - Hot genius free-lance biotech researcher unravels the key component of a radical life-extension gene therapy that will ensure 300 years of robust life to its recipients. The researcher is murdered shortly after he hides the critical component. His distraught friend has him cryogenically frozen. 150 years later, the researcher is revived by the same major bio-med corporation for which he had originally been working.
Quickly he realizes that their motives are less than altruistic: his modification of the gene therapy is needed to resolve an unforeseen debilitation now creeping up in the recipients of the life-extension process. The recipients, now nearing 125 years off added life, are decompensating into psychotics. The researcher at first tries to remember and reconstruct what he did with the hidden critical component, but stops in disgust when he learns that in the past 150 years the life-extension therapy has been reserved solely for the ultra-affluent and has created an extreme and cruel global gerontocratic elite. He voices his disgust to his corporate minders, who cease being beneficent and show their true colors as trying to gain control of this critical technology in order to control the elites.
In the process of dealing with the corporation, he learns about his murder and begins investigating.As he comes closer to the identity of his murderer, he uncovers a wider conspiracy and is the target of more murder attempts.
He was killed by a friend in 2015. The friend was the CEO of a small bio-gen firm that the researcher was doing the LET work for. The CEO, a biz-head with a genetics academic background, took the researcher’s work and exploited it as his own, in the process growing his small firm into a bio-med powerhouse and him into one of the world’s wealthiest individuals.
The CEO also was the first recipient of the LET and is now 190 years old, but doesn’t look a day over 45. Smart, urbane, ruthless, the CEO used his wealth and position to start the cabal of Ultras. It is a faction of the top 50 smartest and wealthiest people in the world who have ‘ascended from the world’ (faked their demise) and control the global economy with their vast coordinated wealth. Perhaps they will call themselves ‘The Ascended’. We need to decide how the cabal lives. Are they sequestered on a luxurious island compound, or do they live in the open, surgically re-sculpted after each faked death, or do they live in the open.
Also we need to figure out what the world will look and feel like in 150 years.
As the ultras decompensate into psychosis, the CEO orders the researcher to be revived in order to find a cure. The CEO had the researcher’s lab notes decrypted and figured that the he was close if not successful in finding the missing component to stabilize the LET.
Tiberius Syndrome: the decline into cruel psychosis experienced by the ultras, named after the roman emperor Tiberius’ degenerate behavior after he sequestered himself on Capri.
The ironic twist might be that there is no cure, no stabilization. The psychosis is not the result of the LET alone, but also due in part to the unfettered ego/wills of the ultras. Absolute power corrupts…
________________________________________________________
a brazilian hacking syndicate was subcontracted by a st petersberg crew to run interference on a hit on SWIFT, the global currency clearinghouse notification network. The UniFavela clan was going to run a multi-flank raid. They specialized in fast propagating virii and had created a custom mail-in virus that exploited a few microsoft vulnerabilities that they had discovered and kept mum. Their target was a Latin American PR spokesman listed on the corporate web site for press queries. The PR flak would be just the sleepy guard on the wall for their virus to slip past. 30 minutes after opening an inocuous spoofed email from a French e-trade publication requesting clarification on the SWIFT-Indentrus partnership. the virus would port scan and map its entire site LAN, salmoning its way up the router paths till it found the deep waters of the main corporate campus network in Brussels. Shortly, the internal LAN at Brussels would be suffering switch and router buffer overflows and traffic would gasp, ack, and sputter. UniFavela would then towel whip out a vanilla DDOS on the main company web site, any INTERNIC-registered addresses, and any other system in the IP block reserved for SWIFT that had previously port scanned as interesting, or ,even, as nothing. Mongols charging the village gates and tossing flaming torches on thatched roofs. IT Operations would be running to and fro, trying to figure out the internal bandwidth crunch and if there was a bleedout causing the external net problems.
____________________________________________________________
The Post-Human Story of Minos:
the CEO of a powerful commercial combine is bore an illegitimate son by his indiscreet wife in retaliation for his own dalliances. the son has a hideous deformity but is fantastically brilliant - brilliant enough for the father overcome his own repulsion of the child - as a bastard and a freak. the father sequesters the child in an elaborate virtual domain. the child, a hacker savant, is used to breach competitor nets. but as his power in the digital realm expands, the child transforms into the tyrant-monster. using the nets, he lashes out at people who have caused him pain, then evolves into enjoying the taste of terror and fear. He becomes the Minotaur.
____________________________________________________________
'there was a mad scramble amongst all the big spook governments, dark side corporations, and the privacy maccabees once it was determined that quantum computation had left the tidal pool of academia, grown legs and air-breathing lungs, and was headed for the nat sec intel highlands. all previous encryption models were rendered obsolete, and worse, exposed. QC became an undefiable xray spotlight, laying bare any encrypted secret with a ease of opening a mathematical candy wrapper. And for a while it swung the advantage back to the state in the digital Boer War against the freecon partisans.'
____________________________________________________________
The Oort, to the Intras, looked as one people. Extra-stellar hillbillies, ekeing out a subsistance existence on extracted organics from the frozen crud comets and other planetesimals of the Oort Cloud that slung around the solar system in a 1K AU circuit. To the Oort there was no Oort. Each station, each kampong was distinct and seperate. Seperate dialects, traditions, norms, goals. Some were scientific collectives, some were tired mining operations, some were intense sectarian cults - they shared little between themselves beyond necessary trade links for scarce commodities.
---
A young prince is disgraced in an internal court scandal and sent into a quasi-exile on a worthless mission. On his travels he builds the wisdom and learns the skills necessary to be a just and effective leader.
His exile was a gambit by his patriarch to remove Genji from the arena of pointless court intrigues and develop him as a real leader. The patriarch dispatched a team of loyal praetorians to discreetly follow and protect Genji on his odyssey.
Genji was sent as an emissary to the Oort system. He must pass through the Martian-Saturnine corridor, populated with industrial trading guilds and their private militias.
----
Genealogy becomes paramount in a closed culture; hierarchy by heredity. Reference the roman patrician class’ death-grip obsession with lineage, or the medieval Japanese imperial court’s strict intra-elite caste system.
But in an era of extreme genetic engineering, how can bloodlines retain their importance? Perhaps this is the wrong question. Perhaps in an era of extreme genetic engineering, authentic bloodlines can only retain their importance. The longevity of an unchanged gene line demonstrates success in evolutionary competition. Over time however, the fitness of a rigidly enforced and ‘sequestered’ gene line will degrade. Consider the hemophilia of the European royal strata.
I would not want the imperial court of the inner system to be pure blue bloods, eschewing genetic manipulation. Rather I would have them take the opposite tack – and embrace genetic engineering in the pursuit of perfecting particular socially valued or distinctive attributes; a roman nose, elongated refined fingers, even the possession of certain ‘noble’ afflictions (for ex., the aforementioned hemophilia as a sign of noble lineage).
The elites should pursue genealogy with the same passion and gusto as horse breeders; studs and mares and percentages of bloodlines, enforced and suppressed gene expressions, surrogates, and gene modes des saisons.
__________________________________________________________
a bum finds a the wallet and keys of a man who jumped from a bridge
he goes to his townhouse to find something to eat or steal
is impressed and overwhelmed with the man's townhouse
showers, eats, gets cleaned up, finds some clothes
is ready to leave when he helps a woman wrestling with groceries at her door
she thanks him, but looks stunned.
‘are you the man in #560? umm..i have lived here for 3 years and have never actually seen you. you seem to leave so early in the morning and get
home so late and keep to yourself.’
they spend 30 minutes talking, having a generally warm friendly encounter.
‘well, I am so glad to have finally met you. Hope to see you soon.’ As she closes her door, the bum turns to leave but pauses and thinks for a moment, then goes back into the man's townhouse
he pours through the man's papers and keepsakes and learns that the man has no family that he speaks with, no friends, lives off a well-endowed trust fund
and
the bum moves in and takes over the mans identity
he brings warmth and sincerity to the man's identity
what makes a hermit tick? what lengths do they go to to remove themselves from society? does it become a game to avoid contact, trying to become a shadow, a phantom? does society dissolve away as a mental force in their thoughts, atrophy away or does it become an amputated impression?
what divsion line stands between a hermit and convict in solitary? the hermit, by and large, chooses their isolation, the convict has it enforced upon them. at what point does the human need for society or socialization collapse? is there anything left that we can inspect and evaluate? a hermit, however, is able to maintain walls against the Great Other, which would imply that they are seeking refuge from the world. a schizo or an autistic will be physically surrounded by others but unable or incapable of making contact.
when does the will to contact die? what is left over? do humans require contact to retain our humanity? can you love and sacrifice in a vacuum?
what defines humanity? oooh, a big question...
___________________________________________________________________
genetic engineering will continue to deconstruct the human species
there will be catastrophic disasters: gene sequence specific viruses engineered to attack 'types' of people. Der Genkampf
petroleum will be replaced- hydrogen-powered locomotion and green power (in the wealthy states). the poor states will continue to be held hostage to oil politics
(cultures and civilizations do not move forward uneringly. they spasticly jerk forward and fro, in clumps andgrains, never ever as a lemming death drive.)
developed economies will be netized. a new state structure will be needed to manage and dsitribute resources. the corporate structure, the commercial backbone of the capitalist democracy, will replace the republic. it is flexible to markets and political forces, insistent on accountability, it provides a sufficient compromise between individual representation and republican government. they will begin their political evolution as projects in community development. assurances of an educated workforce by charter education. assurances of uninterrupted utilities by running their own power/water etc. net-based marketplaces create corporate agoras. employees are in fact de facto citizens of the corporation. citizenship, or regular employment, will be a reward for merit, stock shares will count towards suffrage.
great corporate collectives will arise. housing, education, security...all the needs of the middle class will be absorbed in the corporate state. the tradtional state will cede roles and responsibilities to the corporate state as their resources dwindle. a few isolated violent reactions (military or legal)by the republics against the corporate states, but they will fail over time. against, or more so, in conjunction with the homogenized corporatsists wil be the diasporae, non-corporates will glom to other modes of networked alignment, ethnic allegiance will become stronger over time - as the chinese, indian, and jewish disporaestrengthen as a formula for a successful competition against/with the corporates.
the american state, succored by its overwhelming techo-military supremancy, loses its mission, its vision - substitutes will to dominate for will to excel - and falls into the deep narcotic, insulated slumber of the unassailable. GE, nano, and the banknote net weaken the mythic cohesion of the american spirit. we are no longer united by common experience (mass-mediated or otherwise) the promise of science to make us stronger, smarter, near immortal is held like a manifest destiny or a divine IOU for services rendered to humanity.
Does computational photography in a smart phone obviate fast glass on ILCs?
Happy Food Friday!
This may be the last one from Croatia. some of the fans will be sighing in relief. There are still a few pretty pictures in the backlog, maybe I’ll get desparate and raid the fridge in coming months.
This visualization shows early test renderings of a global computational model of Earth's atmosphere based on data from NASA's Goddard Earth Observing System Model, Version 5 (GEOS-5). This particular run, called Nature Run 2, was run on a supercomputer, spanned 2 years of simulation time at 30 minute intervals, and produced Petabytes of output.
The visualization spans a little more than 7 days of simulation time which is 354 time steps. The time period was chosen because a simulated category-4 typhoon developed off the coast of China.
The 7 day period is repeated several times during the course of the visualization.
Credit: NASA's Scientific Visualization Studio
Read more or download here: svs.gsfc.nasa.gov/goto?4180
NASA Goddard Space Flight Center enables NASA’s mission through four scientific endeavors: Earth Science, Heliophysics, Solar System Exploration, and Astrophysics. Goddard plays a leading role in NASA’s accomplishments by contributing compelling scientific knowledge to advance the Agency’s mission.
Follow us on Twitter
Like us on Facebook
Find us on Instagram
Computational feature (in camera focus stacking) was used to create the DOF to render both subjects in focus. Olympus OM-1 is amazing...!!!!
Abstract composition of the Engineering, Environment and Computing Building at Coventry University. A riot of geometric futuristic shapes.
The ability to fake a long exposure on your phone with no ND filter or tripod is awesome, but the reality of it is... good but not great.
Computationally Challenge, Week 19 - Visual Weight and Balance, Depth of Field
Pentax SMC 50mm f/1.2, taken at f/1.2
Explored 5/14/2025! Thanks for all the views, faves and kind comments!
Modeled in Blender 2.79 and rendered using the LuxCoreRender 2.1 render engine.
This surface represents a surface of constant potential where source terms have been randomly allocated a position and size within a cube. Each source term is given the equation:
f(x,y,z) = e^(-a * ((x-xi)^2 + (y-yi)^2 + (z-zi)^2))
where xi, yi, zi is a random coordinate with the cube and a is a random positive constant.
How do you compute the volume of a cat? Dunking it in water doesn't work-- you get the volume of the rat-like creature that lives inside the cat; much like the feeble alien within a Dalek. (And, if your answer had anything to do with contour integrals, get real.) Here is a method that works: Using successive approximation, determine the smallest box that the cat will fully enclose itself in.
This cat is approximately 648 cubic inches in volume.
Related blog entry here
Dr Gearenstein's new & improved Computation Terminal MK III, now 100% Tesla electric, no steam power required.
A dark night scene taken with a Pixel 4A using the "Night Sight" mode. This is "computational photography," Handheld mobile phone, several second exposure and multiple exposures that are processed to render what the computer (AI) believes the scene looks like. Not perfect, but far better than anything I could get without a tripod.
Computational domes. The design is generated with shape grammars and the construction is adapted with a catenary-simulation. Scripted in Processing.
If academic disciplines are playing card suits, then Computer Science is the joker in the pack
www.cdyf.me/computing?q=joker#joker
Public domain image of the Jolly Joker, a vintage Masenghini Italian playing card via Wikimedia Commons w.wiki/35EW
Congratulations to Intel on their acquisition of Nervana. This photo is from the last board meeting at our offices; the Nervana founders — from right to left: Naveen Rao, Amir Khosrowshahi and Arjun Bansal — pondered where on the wall they may fall during M&A negotiations.
We are now free to share some of our perspectives on the company and its mission to accelerate the future with custom chips for deep learning.
I’ll share a recap of the Nervana story, from an investor’s perspective, and try to explain why machine learning is of fundamental importance to every business over time. In short, I think the application of iterative algorithms (e.g., machine learning, directed evolution, generative design) to build complex systems is the most powerful advance in engineering since the Scientific Method. Machine learning allows us to build software solutions that exceed human understanding, and shows us how AI can innervate every industry.
By crude analogy, Nervana is recapitulating the evolutionary history of the human brain within computing — moving from the logical constructs of the reptilian brain to the cortical constructs of the human brain, with massive arrays of distributed memory and iterative learning algorithms.
Not surprisingly, the founders integrated experiences in neuroscience, distributed computing, and networking — a delightful mélange for tackling cognitive computing. Ali Partovi, an advisor to Nervana, introduced us to the company.
We were impressed with the founding team and we had a prepared mind to share their enthusiasm for the future of deep learning. Part of that prepared mind dates back to 1989, when I started a PhD in EE focusing on how to accelerate neural networks by mapping them to parallel processing computers. Fast forward 25 years, and the nomenclature has shifted to machine learning and the deep learning subset, and I chose it as the top tech trend of 2013 at the Churchill Club VC debate (video). We were also seeing the powerful application of deep learning and directed evolution across our portfolio, from molecular design to image recognition to cancer research to autonomous driving.
All of these companies were deploying these simulated neural networks on traditional compute clusters. Some were realizing huge advantages by porting their code to GPUs; these specialized processors originally designed for rapid rendering of computer graphics have many more computational cores than a traditional CPU, a baby step toward a cortical architecture. I first saw them being used for cortical simulations in 2007. But by the time of Nervana’s founding in 2014, some (e.g., Microsoft’s and Google’s search teams) were exploring FPGA chips for their even finer-grained arrays of customizable logic blocks. Custom silicon that could scale beyond any of these approaches seemed like the natural next step. Here is a page from Nervana’s original business plan (Fig. 1 in comments below).
The march to specialized silicon, from CPU to GPU to FPGA to ASIC, had played out similarly for Bitcoin miners, with each step toward specialized silicon obsoleting the predecessors. When we spoke to Amazon, Google, Baidu, and Microsoft in our due diligence, we found a much broader application of deep learning within these companies than we could have imagined prior, from product positioning to supply chain management.
Machine learning is central to almost everything that Google does. And through that lens, their acquisition, and new product strategies make sense; they are not traditional product line extensions, but a process expansion of machine leaning (more on that later). They are not just playing games of Go for the fun of it. Recently, Google switched their core search algorithms to deep learning, and they used Deep Mind to cut data center cooling costs by a whopping 40%.
The advances in deep learning are domain independent. Google can hire and acquire talent and delight in their passionate pursuit of game playing or robotics. These efforts help Google build a better brain. The brain can learn many things. It is like a newborn human; it has the capacity to learn any of the languages of the world, but based on training exposure, it will only learn a few. Similarly, a synthetic neural network can learn many things.
Google can let the Brain team find cats on the Internet and play a great game of Go. The process advances they make in building a better brain (or in this case, a better learning machine) can then be turned to ad matching, a task that does not inspire the best and the brightest to come work for Google.
The domain independence of deep learning has profound implications on labor markets and business strategy. The locus of learning shifts from end products to the process of their creation. Artifact engineering becomes more like parenting than programming. But more on that later; back to the Nervana story.
Our investment thesis for the Series A revolved around some universal tenets: a great group of people pursuing a product vision unlike anything we had seen before. The semiconductor sector was not crowded with investor interest. AI was not yet on many venture firms’ sectors of interest. We also shared with the team that we could envision secondary benefits from discovering the customers. Learning about the cutting edge of deep learning applications and the startups exploring the frontiers of the unknown held a certain appeal for me. And sure enough, there were patterns in customer interest, from an early flurry in medical imaging of all kinds to a recent explosion of interest in the automotive sector after Tesla’s Autopilot feature went live. The auto industry collectively rushed to catch up.
Soon after we led the Series A on August 8, 2014, I found myself moderating a deep learning panel at Stanford with Nervana CEO Naveen Rao.
I opened with an introduction to deep learning and why it has exploded in the past four years (video primer). I ended with some common patterns in the power and inscrutability of artifacts built with iterative algorithms. We see this in biology, cellular automata, genetic programming, machine learning and neural networks.
There is no mathematical shortcut for the decomposition of a neural network or genetic program, no way to “reverse evolve” with the ease that we can reverse engineer the artifacts of purposeful design.
The beauty of compounding iterative algorithms — evolution, fractals, organic growth, art — derives from their irreducibility. (More from my Google Tech Talk and MIT Tech Review)
Year 1. 2015
Nervana adds remarkable engineering talent, a key strategy of the first mover. One of the engineers figures out how to rework the undocumented firmware of NVIDIA GPUs so that they run deep learning algorithms faster than off-the-shelf GPUs or anything else Facebook could find. Matt Ocko preempted the second venture round of the company, and he brought the collective learning of the Data Collective to the board.
Year 2. 2016 Happy 2nd Birthday Nervana!
The company is heads down on chip development. They share some technical details (flexpoint arithmetic optimized for matrix multiplies and 32GB of stacked 3D memory on chip) that gives them 55 trillion operations per second on their forthcoming chip, and multiple high-speed interconnects (as typically seen in the networking industry) for ganging a matrix of chips together into unprecedented compute fabrics. 10x made manifest. See Fig. 2 below.
And then Intel came knocking.
With the most advanced production fab in the world and a healthy desire to regain the mantle of leading the future of Moore’s Law, the combination was hard to resist. Intel vice president Jason Waxman told Recode that the shift to artificial intelligence could dwarf the move to cloud computing. “I firmly believe this is not only the next wave but something that will dwarf the last wave.” But we had to put on our wizard hats to negotiate with giants.
The deep learning and AI sector have heated up in labor markets to relatively unprecedented levels. Large companies are recently paying $6–10 million per engineer for talent acquisitions, and $4–5M per head for pre-product startups still in academia. For the Masters students in a certain Stanford lab, they averaged $500K/yr for their first job offer at graduation. We witnessed an academic turn down a million dollar signing bonus because they got a better offer.
Why so hot?
The deep learning techniques, while relatively easy to learn, are quite foreign to traditional engineering modalities. It takes a different mindset and a relaxation of the presumption of control. The practitioners are like magi, sequestered from the rest of a typical engineering process. The artifacts of their creation are isolated blocks of functionality defined by their interfaces. They are like blocks of magic handed to other parts of a traditional organization. (This carries over to the customers too; just about any product that you experience in the next five years that seems like magic will almost certainly be built by these algorithms).
And remember that these “brain builders” could join any industry. They can ply their trade in any domain. When we were building the deep learning team at Human Longevity Inc. (HLI), we hired the engineering lead from the Google’s Translate team. Franz Och pioneered Google’s better-than-human translation service not by studying linguistics, grammar, or even speaking the languages being translated. He focused on building the brain that could learn the job from countless documents already translated by humans (UN transcripts in particular). When he came to HLI, he cared about the mission, but knew nothing about cancer and the genome. The learning machines can find the complex patterns across the genome. In short, the deep learning expertise is fungible, and there are a burgeoning number of companies hiring and competing across industry lines.
And it is an ever-widening set of industries undergoing transformation, from automotive to agriculture, healthcare to financial services. We saw this explosion in the Nervana customer pipeline. And we see it across the DFJ portfolio, especially in our newer investments. Here are some examples:
• Learning chemistry and drug discovery: Here is a visualization of the search space of candidates for a treatment for Ebola; it generated the lead molecule for animal trials. Atomwise summarizes: “When we examine different neurons on the network we see something new: AtomNet has learned to recognize essential chemical groups like hydrogen bonding, aromaticity, and single-bonded carbons. Critically, no human ever taught AtomNet the building blocks of organic chemistry. AtomNet discovered them itself by studying vast quantities of target and ligand data. The patterns it independently observed are so foundational that medicinal chemists often think about them, and they are studied in academic courses. Put simply, AtomNet is teaching itself college chemistry.”
• Designing new microbial life for better materials: Zymergen uses machine learning to predict the combination of genetic modifications that will optimize product yield for their customers. They are amassing one of the largest data sets about microbial design and performance, which enables them to train machine learning algorithms that make search predictions with increasing precision. Genomatica had great success in pathway optimization using directed evolution, a physical variant of an iterative optimization algorithm.
• Discovery and change detection in satellite imagery: Planet and Mapbox. Planet is now producing so much imagery that humans can’t actually look at each picture it takes. Soon, they will image every meter of the Earth every day. From a few training examples, a convolutional neural net can find similar examples globally — like all new housing starts, all depleted reservoirs, all current deforestation, or car counts for all retail parking lots.
• Automated driving & robotics: Tesla, Zoox, SpaceX, Rethink Robotics, etc.
• Visual classification: From e-commerce to drones to security cameras and more. Imagen is using deep learning to radically improve medical image analysis, starting with radiology.
• Cybersecurity: When protecting endpoint computing & IOT devices from the most advanced cyberthreats, AI-powered Cylance is proving to be a far superior and adaptive approach versus older signature-based antivirus solutions.
• Financial risk assessment: Avant and Prosper use machine learning to improve credit verification and merge traditional and non-traditional data sources during the underwriting process.
• And now for something completely different: quantum computing. For a wormhole peek into the near future, our quantum computing company, D-Wave Systems, powered a 100,000,000x speedup in a demonstration benchmark for Google, a company that has used D-Wave quantum computers for over a decade now on machine learning applications.
So where will this take us?
Neural networks had their early success in speech recognition in the 90’s. In 2012, the deep learning variant dominated the ImageNet competitions, and visual processing can now be better done by machine than human in many domains (like pathology, radiology and other medical image classification tasks). DARPA has research programs to do better than a dog’s nose in olfaction.
We are starting the development of our artificial brains in the sensory cortex, much like an infant coming into the world. Even within these systems, like vision, the deep learning network starts with similar low level constructs (like edge-detection) as foundations for higher level constructs like facial forms, and ultimately, finding cats on the internet with self-taught learning.
But the artificial brains need not limit themselves to the human senses. With the internet of things, we are creating a sensory nervous system on the planet, with countless sensors and data collecting proliferating across the planet. All of this “big data” would be a big headache but for machine learning to find patterns in it all and make it actionable. So, not only are we transcending human intelligence with multitudes of dedicated intelligences, we are transcending our sensory perception.
And it need not stop there. It is precisely by these iterative algorithms that human intelligence arose from primitive antecedents. While biological evolution was slow, it provides an existence proof of the process, now vastly accelerated in the artificial domain. It shifts the debate from the realm of the possible to the likely timeline ahead.
Let me end with the closing chapter in Danny Hillis’ CS book The Pattern on the Stone: “We will not engineer an artificial intelligence; rather we will set up the right conditions under which an intelligence can emerge. The greatest achievement of our technology may well be creation of tools that allow us to go beyond engineering — that allow us to create more than we can understand.”
-----
Here is some early press:
Xconomy(most in-depth), MIT Tech Review, Re/Code, Forbes, WSJ, Fortune.
Computational domes. The design is generated with shape grammars and the construction is adapted with a catenary-simulation. Scripted in Processing.
Timnit Gebru wants to replace the U.S. Census (which costs $1B/year to implement) by simply analyzing the cars seen in Google Street View images.
After processing 22 million observed cars, she found some fascinating things, like the predictive power of "the sedan/truck ratio" for political party. Republicans sure like trucks! More findings in the comments below.
From the AI in Fintech Forum today at Stanford ICME.
This is a Computational Fluid Dynamics (CFD) computer-generated model of the Space Shuttle during re-entry. CFD has supplanted wind tunnels for many evaluations of aircraft. As computing power increases and computer models become more sophisticated, CFD will become an increasingly more powerful tool for aeronautics research.
Credit: NASA
Image Number: L-1993-03205
Date: April 1, 1993
I created this after teaching my Creative Computation 1 students how to data bend using a variety of ways (including Antonio Roberts excellent tutorial on databending in Audacity www.hellocatfood.com/databending-using-audacity/), this was done using the wah wah filter in Audacity. I bent 3 files and merged them in photoshop to make the final.
These components perform key computations for Tide Predicting Machine No. 2, a special purpose mechanical analog computer for predicting the height and time of high and low tides. The tide prediction formula implemented by the machine includes the addition of a series of cosine terms. The triangular metal pieces are part of slotted yoke cranks which convert circular motion to a vertical motion that traces a sinusoid. Each slotted yoke crank is connected by a shaft to a pulley, which causes the pulley to follow the sinusoidal motion. A chain going over and under pulleys sums each of their deflections to compute the tide. Along the top of the photo, connecting shafts drive slotted yoke cranks on both sides of the machine.
The U.S. government used Tide Predicting Machine No. 2 from 1910 to 1965 to predict tides for ports around the world. The machine, also known as “Old Brass Brains,” uses an intricate arrangement of gears, pulleys, chains, slides, and other mechanical components to perform the computations.
A person using the machine would require 2-3 days to compute a year’s tides at one location. A person performing the same calculations by hand would require hundreds of days to perform the work. The machine is 10.8 feet (3.3 m) long, 6.2 feet (1.9 m) high, and 2.0 feet (0.61 m) wide and weighs approximately 2,500 pounds (1134 kg). The operator powers the machine with a hand crank.
The National Oceanic and Atmospheric Administration (NOAA) occasionally displays the machine at its facility in Silver Spring, Maryland.