IBM’s resistive computing could massively accelerate AI — and get us closer to Asimov’s Positronic Brain

April 23, 2016

cq5dam.web.1280.1280

With the recent rapid advances in machine learning has come a renaissance for neural networks — computer software that solves problems a little bit like a human brain, by employing a complex process of pattern-matching distributed across many virtual nodes, or “neurons.” Modern compute power has enabled neural networks to recognize images, speech, and faces, as well as to pilot self-driving cars, and win at Go and Jeopardy. Most computer scientists think that is only the beginning of what will ultimately be possible. Unfortunately, the hardware we use to train and run neural networks looks almost nothing like their architecture. That means it can take days or even weeks to train a neural network to solve a problem — even on a compute cluster — and then require a large amount of power to solve the problem once they’re trained.

Neuromorphic computing may be key to advancing AI

Researchers at IBM aim to change all that, by perfecting another technology that, like neural networks, first appeared decades ago. Loosely called resistive computing, the concept is to have compute units that are analog in nature, small in substance, and can retain their history so they can learn during the training process. Accelerating neural networks with hardware isn’t new to IBM. It recently announced the sale of some of its TrueNorth chips to Lawrence National Labs for AI research. TrueNorth’s design is neuromorphic, meaning that the chips roughly approximate the brain’s architecture of neurons and synapses. Despite its slow clock rate of 1 KHz, TrueNorth can run neural networks very efficiently because of its million tiny processing units that each emulate a neuron.

Until now, though, neural network accelerators like TrueNorth have been limited to the problem-solving portion of deploying a neural network. Training — the painstaking process of letting the system grade itself on a test data set, and then tweaking parameters (called weights) until it achieves success — still needs to be done on traditional computers. Moving from CPUs to GPUs and custom silicon has increased performance and reduced the power consumption required, but the process is still expensive and time consuming. That is where new work by IBM researchers Tayfun Gokmen and Yuri Vlasov comes in. They propose a new chip architecture, using resistive computing to create tiles of millions of Resistive Processing Units (RPUs), which can be used for both training and running neural networks.

Using Resistive Computing to break the neural network training bottleneck

Deep neural networks have at least one hidden layer, and often hundreds. That makes them expensive to emulate on traditional hardware.Resistive Computing is a large topic, but roughly speaking, in the IBM design each small processing unit (RPU) mimics a synapse in the brain. It receives a variety of analog inputs — in the form of voltages — and based on its past “experience” uses a weighted function of them to decide what result to pass along to the next set of compute elements. Synapses have a bewildering, and not-yet totally understood layout in the brain, but chips with resistive elements tend to have them neatly organized in two-dimensional arrays. For example, IBM’s recent work shows how it is possible to organize them in 4,096-by-4,096 arrays.

Because resistive compute units are specialized (compared with a CPU or GPU core), and don’t need to either convert analog to digital information, or access memory other than their own, they can be fast and consume little power. So, in theory, a complex neural network — like the ones used to recognize road signs in a self-driving car, for example — can be directly modeled by dedicating a resistive compute element to each of the software-described nodes. However, because RPUs are imprecise — due to their analog nature and a certain amount of noise in their circuitry — any algorithm run on them needs to be made resistant to the imprecision inherent in resistive computing elements.

Traditional neural network algorithms — both for execution and training — have been written assuming high-precision digital processing units that could easily call on any needed memory values. Rewriting them so that each local node can execute largely on its own, and be imprecise, but produce a result that is still sufficiently accurate, required a lot of software innovation.

For these new software algorithms to work at scale, advances were also needed in hardware. Existing technologies weren’t adequate to create “synapses” that could be packed together closely enough, and operate with low power in a noisy environment, to make resistive processing a practical alternative to existing approaches. Runtime execution happened first, with the logic for training a neural net on a hybrid resistive computer not developed until 2014. At the time, researchers at the University of Pittsburg and Tsinghua University claimed that such a solution could result in a 3-to-4-order-of-magnitude gain in power efficiency at the cost of only about 5% in accuracy.

IBM researchers claim an RPU-based design will be massively more efficient for neural network applications

IBM researchers claim an RPU-based design will be massively more efficient for neural network applications, shown in this Table from their paper

Moving from execution to training

This new work from IBM pushes the use of resistive computing even further, postulating a system where almost all computation is done on RPUs, with traditional circuitry only needed for support functions and input and output. This innovation relies on combining a version of a neural network training algorithm that can run on an RPU-based architecture with a hardware specification for an RPU that could run it.

As far as putting the ideas into practice, so far resistive compute has been mostly a theoretical construct. The first resistive memory (RRAM) became available for prototyping in 2012, and isn’t expected to be a mainstream product for several more years. And those chips, while they will help scale memory systems, and show the viability of using resistive technology in computing, don’t address the issue of synapse-like processing.

If RPUs can be built, the sky is the limit

The RPU design proposed is expected to accommodate a variety of deep neural network (DNN) architectures, including fully-connected and convolutional, which makes them potentially useful across nearly the entire spectrum of neural network applications. Using existing CMOS technology, and assuming RPUs in 4,096-by-4,096-element tiles with an 80-nanosecond cycle time, one of these tiles would be able to execute about 51 GigaOps per second, using a minuscule amount of power. A chip with 100 tiles and a single complementary CPU core could handle a network with up to 16 billion weights while consuming only 22 watts (only two of which are actually from the RPUs — the rest is from the CPU core needed to help get data in and out of the chip and provide overall control).

That is a staggering number compared to what is possible when chugging data through the relatively lesser number of cores in even a GPU (think about 16 million compute elements, compared with a few thousand). Using chips densely packed with these RPU tiles, the researchers claim that, once built, a resistive-computing-based AI system can achieve performance improvements of up to 30,000 times compared with current architectures, all with a power efficiency of 84,000 GigaOps per-second per-watt. If this becomes a reality, we could be on our way to realizing Isaac Asimov’s fantasy vision of the robotic Positronic brain.

IBM’s resistive computing could massively accelerate AI — and get us closer to Asimov’s Positronic Brain

Mayo Clinic Taps Silicon Valley to Help People Age Gracefully

April 23, 2016

gettyimages-169759918-crop

Could there be a recipe for a longer, healthier life?

Pop a pill and live a long, healthy life. It might not be quite that easy yet, but researchers at Mayo Clinic believe they have found a cell that could hold the secret to aging extra gracefully.

Their research, published in the journal Nature Wednesday, helped patients live longer, healthier lives. The only catch is their patients are mice. But the researchers believe they could someday translate it into a recipe for human longevity, too.

In fact, the research has been so convincing that Mayo Clinic invested in Unity Biotechnology, a San Francisco-based startup built around the researchers’ approach. Other investors in the company include ARCH Venture Partners, Venrock, and Chinese WuXi, and the study’s lead author Jan van Deursen is listed as a Unity co-founder.

The anti-aging method works like this: scientists inject the mice with a drug that pushes out toxic, worn-out cells called “senescent cells.” The senescent cells are old and stressed and don’t behave properly anymore. Instead, they “litter the body with aging” as van Deursen puts it.

In a future brave new world will it be possible to live forever?

April 23, 2016

transhumanism

January is a month for renewal and for change. Many of us have been gifted shiny new fitness trackers, treated ourselves to some new gadget or other, or upgraded to the latest smartphone. As we huff and puff our way out of the season of excess we find ourselves wishing we could trade in our overindulged bodies for the latest model.

The reality is that, even with the best of care, the human body eventually ceases to function but if I can upgrade my smartphone, why can’t I upgrade myself? Using technology, is it not possible to live forever(ish)?

After all, humans have been “upgrading” themselves in various ways for centuries. The invention of writing allowed us to offload memories, suits of armour made the body invincible to spears, eyeglasses gave us perfect 20/20 vision, the list goes on.

This is something that designer and author Natasha Vita-More has been thinking about for a long time. In 1983 she wrote The Transhumanist Manifesto, setting out her vision for a future where technology can lead to “radical life extension” – if not living forever, then living for a lot longer than is currently possible.

Vita-More has also designed a prototype whole body prosthetic she calls Primo PostHuman. This is a hypothetical artificial body that could replace our own and into which we could, in theory, upload our consciousness. This is more in the realm of living forever but is a concept as distant to us as Leonardo da Vinci’s sketch of a flying machine was to 15th century Europeans.

Even so, while the replacement body seems much closer to science fiction than science, recent advances in robotics and prosthetics have not only given us artificial arms that can detect pressure and temperature but limbs that can be controlled by thoughts using a brain-computer interface.

As a transhumanist, Vita-More is excited by these scientific developments. She defines a transhumanist to be “a person who wants to engage with technology, extend the human lifespan, intervene with the disease of aging, and wants to look critically at all of these things”.

Transhumanism, she explains, looks at not just augmenting or bypassing the frailties of the human body but also improving intelligence, eradicating diseases and disabilities, and even equipping us with greater empathy.

“The goal is to stay alive as long as possible, as healthy as possible, with greater consciousness or humaneness. No-one wants to stay alive drooling in a wheelchair,” she adds.

Who wouldn’t want to be smarter, stronger, healthier and kinder? What could possibly go wrong?

A lot, says Dr Fiachra O’Brolcháin, a Marie Curie/Assistid Research Fellow at the Institute of Ethics, Dublin City University whose research involves the ethics of technology.

Take for example being taller than average: this correlates with above average income so it is a desirable trait. But if medical technology allowed for parents to choose a taller than average child, then this could lead to a “height race”, where each generation becomes taller and taller, he explains.

“Similarly, depending on the society, even non-homophobic people might select against having gay children (assuming this were possible) if they thought this would be a disadvantage. We might find ourselves inaugurating an era of ‘liberal eugenics’, in which future generations are created according to consumer choice.”

Then there is the problem of affordability. Most of us do not have the financial means to acquire the latest cutting-edge tech until prices drop and it becomes mainstream. Imagine a future where only the rich could access human enhancements, live long lives and avoid health problems.

Elysium, starring Matt Damon, takes this idea to its most extreme, leading to a scenario similar to what O’Brolcháin describes as “an unbridgeable divide between the enhanced and the unenhanced”.

Despite the hyper focus on these technological enhancements that come with real risks and ethical dilemmas, the transhumanist movement also seems to be about kicking back against – or at least questioning – what society expects of you.

“There’s a certain parameter of what is normal or natural. There’s a certain parameter of what one is supposed to be,” says Vita-More.

“You’re supposed to go to school at a certain age, get married at a certain age, produce children, retire and grow old. You’re supposed to live until you are 80, be happy, die and make way for the young.”

Vita-More sees technology as freeing us from these societal and biological constraints. Why can’t we choose who we are beyond the body we were born with? Scholars on the sociology of the early Web showed that Cyberspace became a place for this precise form of expression. Maybe technology will continue to provide a platform for this reinvention of what it is to be human.

Maybe, where we’re going, we won’t need bodies.

Digital heaven

Nell Watson’s job is to think about the future and she says: “I often wonder if, since we could be digitised from the inside out – not in the next 10 years but sometime in this century – we could create a kind of digital heaven or playground where our minds will be uploaded and we could live with our friends and family away from the perils of the physical world.

“It wouldn’t really matter if our bodies suddenly stopped functioning, it wouldn’t be the end of the world. What really matters is that we could still live on.”

In other words you could simply upload to a new, perhaps synthetic, body.

As a futurist with Singularity University (SU), a Silicon Valley-based corporation that is part university, part business incubator, Watson, in her own words, is “someone who looks at the world today and projects into the future; who tries to figure out what current trends mean in terms of the future of technology, society and how these two things intermingle”.

She talks about existing technologies that are already changing our bodies and our minds: “There are experiments using DNA origami. It’s a new technique that came out a few years ago and uses the natural folding abilities of DNA to create little Lego blocks out of DNA on a tiny, tiny scale. You can create logic gates – the basic components of computers – out of these things.

“These are being used experimentally today to create nanobots that can go inside the bloodstream and destroy leukaemia cells, and in trials they have already cured two people of leukaemia. It is not science fiction: it is fact.”

Nanobots are also able to carry out distributed computing i.e. communicate with each other, inside living things, she says, explaining that this has been done successfully with cockroaches.

Recording everything

“The cockroach essentially has an on-board computer and if you scale this up to humans and optimise it there is no reason why we can’t have our smartphones inside our bodies instead of carrying them around,” she says.

This on-board AI travelling around our bloodstream would act as a co-pilot: seeing what you see, experiencing what you experience, recording everything and maybe even mapping every single neuron in your brain while it’s at it. And with a digitised copy of your brain you (whatever ‘you’ is) could, in theory, be uploaded to the cloud.

Does this mean that we could never be disconnected from the web, ever again? What if your ‘internal smartphone’ is hacked? Could our thoughts be monitored?

Humans have become so dependent on our smartphones and so used to sharing our data with third parties, that this ‘co-pilot’ inside us might be all too readily accepted without deeper consideration.

Already, novel technologies are undermining privacy to an alarming degree, says O’Brolcháin.

“In a world without privacy, there is a great risk of censorship and self-censorship. Ultimately, this affects people’s autonomy – their ability to decide what sort of life they want to lead for themselves, to develop their own conception of the good life.

“This is one of the great ironies of the current wave of technologies – they are born of individualistic societies and often defended in the name of individual rights but might create a society that can no longer protect individual autonomy,” he warns.

Okay, so an invincible body and a super brain have their downsides but what about technology that expands our consciousness, making us wiser, nicer, all-round better folks? Could world peace be possible if we enhanced our morality?

“If you take a look at humanity you can see fighting, wars, terrorism, anger. Television shows are full of violence, society places an emphasis on wealth and greed. I think part of the transhumanist scope is [to offset this with] intentional acts of kindness,” says Vita-More, who several times during our interview makes the point that technology alone cannot evolve to make a better world unless humanity evolves alongside.

Vita-More dismisses the notion of enhancement for enhancement’s sake, a nod to the grinder movement of DIY body-hacking, driven mostly by curiosity.

Examples include implanting magnets into the fingertips to detect magnetic waves or sticking an RFID chip into your arm as UK professor Kevin Warwick did, allowing him to pass through security doors with a wave of his hand.

Moral enhancements

Along the same lines as Vita-More’s thinking, O’Brolcháin says “some philosophers argue that moral enhancements will be necessary if enhancements are not to be used for malevolent ends”.

“Moral enhancement may result in people who are less greedy, less aggressive, more concerned with addressing serious global issues like climate change,” he muses.

But the difficulty is deciding on what is moral. After all, he says, the ‘good’ groups like Isis want to promote is vastly at odds with the values of Ireland. So who gets to decide what moral enhancements are developed? Perhaps they will come with the latest internal smartphone upgrade or installed at birth by government.

Technology does make life better and it is an exciting time for robotics, artificial intelligence and nanotechnology. But humans have a long way to go to before we work out how we can co-exist with the future we are building right now.

http://www.irishtimes.com/business/in-a-future-brave-new-world-will-it-be-possible-to-live-forever-1.2498427

The Antipode: Flying from New York to London in 11 minutes

April 23, 2016

Antipode_intro_3562097b

(CNN)Remember the Skreemr, a concept for a supersonic plane that could travel at Mach 10?

Scratch that, there’s now a design for a plane that could cruise from London to New York in 11 minutes, traveling at Mach 24 — that’s 12 times faster than the Concorde!
Charles Bombardier, the industrial designer who came up with both designs, has dubbed this newest concept the Antipode, which he conceived in collaboration with Lunatic Koncepts founder Abhishek Roy.
In theory, it could carry up to ten passengers up to 12,430 miles in under an hour.

How it would work

“I wanted to create an aircraft concept capable of reaching its antipode — or diametrical opposite — as fast as possible,” Bombardier told Forbes.
Did you just daze out? Bear with us a little while longer while we explain the tech that would get this craft flying.
The Antipode’s wings would be fitted with rocket boosters that would propel the aircraft to 40,000 feet, and enable it to reach Mach 5.
Like the Skreemr, the plane would be powered by a scramjet engine.
Unlike conventional jet engines, scramjet engines have virtually no moving parts.
And unlike rockets, scramjet engines would burn oxygen from the atmosphere instead of having to carry heavy tanks full of oxygen.

But is it feasible?

Now, the Skreemr concept got some flack for the challenges presented using some of the same technology.
One big problem was heat.
Objects traveling past Mach 5 can reach upwards of 980 C (1800 F), and there is a limit to the type of materials that can withstand those kinds of temperatures.
The sonic boom is also all but guaranteed when an object breaks the sound barrier, and is a menace in urban areas.
However, Bombardier believes he may have found a solution to both issues.
After the Skreemr concept made the rounds, Bombardier was contacted by Joseph Hazeltine, an engineer at Wyle, which provides technical support to both NASA and the U.S. Department of Defence.
Hazeltine suggested using an aerodynamic technique called long penetration mode, or LPM, which would use a nozzle on the aircraft’s nose to blow out air and cool down the surface temperature, while muffling the noise made from breaking the sound barrier.
Yes, it’s above our heads too. Still, impressive though these techniques all sound, most of the technology in this design is still decades away from seeing the light of day.
Even NASA hasn’t created a stable scramjet yet. The Pentagon came closest, launching a small, unmanned scramjet aircraft in 2013 that hit Mach 5.
Note: There’s been nothing commercial, and nothing approaching anything near the speed Bombardier is suggesting with the Antipode.
The designer himself doesn’t seem too worried that his concept is still decades away.
As he explains in a video on his website:
“It’s all about innovation. Share your idea, and it opens up a door for other designers to build on it.”

Samsung predicts the world 100 years from now

April 23, 2016

 

Hyper-tall skyscrapers, underwater bubble cities, personal home “medi-pods” and civilian colonies on the Moon are all likely to be a reality in a hundred years’ time, according to a report commissioned by Samsung.

A new study commissioned by Samsung paints a vivid picture of our future lives; suggesting the way we live, work and play will change beyond recognition over the course of the next century. The SmartThings Future Living Report was authored by a team of leading academics – including TV presenter and space scientist, Dr Maggie Aderin-Pocock, award-winning futurist architects and lecturers at the University of Westminster, Arthur Mamou-Mani and Toby Burgess, as well as pioneering urbanists Linda Aitken and Els Leclerq.

The report was released to promote SmartThings, a system which allows people to make their home smarter, meaning that at any time and from anywhere it is possible to switch on lights, turn up the thermostat or unlock the back door, all via a simple app or automatically through daily routines – something that might have seemed like science fiction as little as 10 years ago, but today is a reality.

The predictions for how we will live in the future have been brought to life via detailed animated renders, showing a futuristic London where high rise apartments dwarf the Shard, and drone transportation is ubiquitous.

 

future london skyline 2100

 

Many of the predictions were influenced by environmental conditions, with growing populations leading to the development of structures better able to cope with space constraints and diminishing resources. As city space becomes ever more squeezed, we will burrow deeper and build higher with the creation of:

Super skyscrapers: carbon nanotubes and diamond nano-threads will help us create towering megastructures that dwarf today’s skyscrapers

Earth-scrapers: just as we build up, we will also dig down – huge structures will tunnel 25 storeys deep, or more

Underwater cities: using the water itself to create breathable atmospheres and generating hydrogen fuel in the process

Personal flying drones replacing cars: we will travel through “skyways” with our own personal flying drones, some big enough to carry entire homes around the world for holidays

 

Click to enlarge

underwater bubble city 2100

 

As technology develops, Samsung predicts:

3D printing of houses and furniture: we will be able to print exact replicas of large scale structures like houses out of local, recyclable materials so that we really can have all the comforts of home while we are away

Flexible, smart walls and 3D printed Michelin starred meals: smart walls will mean you won’t need to decorate your home – room surfaces will adapt to suit your mood. When it comes to entertaining, there will be no more botched recipes or pizza deliveries – instead we will be downloading dishes from famous chefs that we will tailor to our personal needs. We will be able to 3D-print a banquet or a favourite cake in minutes

Virtual meetings: our working lives will be transformed with the use of holograms which will allow us to attend meetings virtually, without leaving the comfort of our homes

Home medi-pods: stepping into these will confirm if you really are ill, providing a digital diagnosis and supplying medicine or a remote surgeon if needed

Colonisation of space: first the Moon, then Mars, then far beyond into the galaxy

 

future civilian moon colony 2100

 

In addition to looking at how we will live in 100 years’ time, the SmartThings team surveyed 2,000 British adults to pinpoint the predictions the nation thought were the most likely to become a reality in the future. This survey shines a spotlight on the public perception of the future and suggests that building further into the sky, and colonising oceans are believed to be the biggest future trends, as space and resources become scarce. The top ten predictions for future living:

 

1 Virtual work meetings – the ability to work from anywhere and attend meetings remotely via avatars/holograms 48%
2 Commercial flights into space 41%
3 Virtual interior decoration to program your surroundings/LED walls that adapt surroundings to your mood 26%
4 3D printed houses/furniture/food – you can instantly download and print these things at home 25%
5 At-home scanning capsules/pods that can diagnose health problems and administer medicines/treatments 24%
6 Colonising other planets as we use up resources on Earth 19%
7 AI becoming a normal part of daily life – taking over from humans in many industries 18%
8 Giant skyscrapers that house entire cities, built with new super-strong materials 18%
9 At-home hydroponic farms (that don’t require soil) where you can grow your own food 17%
10 Earth-scrapers – parts of cities becoming subterranean, due to space constraints and also to provide further shelter 16%

 

Space Scientist Maggie Aderin-Pocock, who co-authored the report, commented: “Our lives today are almost unrecognisable from those a century ago. The Internet has revolutionised the way we communicate, learn and control our lives. Just 25 years ago, technology like SmartThings would have been inconceivable, yet today, developments like this let us monitor, control and secure our living spaces with the touch of a smartphone. Over the next century we will witness further seismic shifts in the way we live and interact with our surroundings – working on the SmartThings Future Living Report, with a panel of industry experts, has allowed me to explore what these could be.

“We are likely to see the emergence of towering megastructures, as well as sub-aquatic cities and transportation via advanced flying drones – some of which could be strong enough to transport entire houses on holiday.”

 

drone flying house 2100

 

James Monighan, UK Managing Director of Samsung SmartThings, comments as follows: “The smart home revolution will have massively positive implications on how we live. Our homes are becoming smarter – they can now detect the presence of things like people, pets, smoke, humidity, lighting and moisture. And this is just the beginning.

“Just as the technology driving the Internet has spread to smartphones and smart homes, the smart home revolution is destined to spread to larger communities and countries. By simply turning lights and heating off when we don’t use them, we can reduce emissions. By being able to better monitor and secure our homes, we can reduce crime. By better monitoring the habits of aging relatives, we help them to achieve greater independence and a higher quality of life.”

The report has been published to coincide with the announcement that SmartThings will work with hundreds of products, from a wide range of brands – as well as working with all of Samsung’s TVs, refrigerators, washer machines, ovens, and robot vacuum cleaners.

http://www.futuretimeline.net/blog/2016/02/24.htm#.VxuCl0c1aM9