 Your new post is loading...
At the Association for Computing Machinery's Conference on Systems, Programming, Languages and Applications: Software for Humanity (SPLASH), researchers from MIT, the French Alternative Energies and Atomic Energy Commission, and Adobe Research recently presented a new system that automatically produces code optimized for sparse data. Tensor algebra is a powerful tool with applications in machine learning, data analytics, engineering and the physical sciences. Tensors are often sparse and compound operations must frequently be computed in a single kernel for performance and to save memory. Programmers are left to write kernels for every operation of interest, with different mixes of dense and sparse tensors in different formats. The combinations are infinite, which makes it impossible to manually implement and optimize all of them. This work presents for the first time a compiler technique to automatically generate kernels for any compound tensor algebra operation on dense and sparse tensors. The technique is implemented in a C++ library called taco. Its performance is competitive with best-in-class hand-optimized kernels in popular libraries, while supporting far more tensor operations.
IBM Research announced Tuesday (Oct. 24, 2017) that its scientists have developed the first “in-memory computing” or “computational memory” computer system architecture, which is expected to yield yield 200x improvements in computer speed and energy efficiency — enabling ultra-dense, low-power, massively parallel computing systems. Their concept is to use one device (such as phase change memory or PCM*) for both storing and processing information. That design would replace the conventional “von Neumann” computer architecture, used in standard desktop computers, laptops, and cellphones, which splits computation and memory into two different devices. That requires moving data back and forth between memory and the computing unit, making them slower and less energy-efficient. The researchers used PCM devices made from a germanium antimony telluride alloy, which is stacked and sandwiched between two electrodes. When the scientists apply a tiny electric current to the material, they heat it, which alters its state from amorphous (with a disordered atomic arrangement) to crystalline (with an ordered atomic configuration). The IBM researchers have used the crystallization dynamics to perform computation in memory. The researchers believe this new prototype technology will enable ultra-dense, low-power, and massively parallel computing systems that are especially useful for AI applications. The researchers tested the new architecture using an unsupervised machine-learning algorithm running on one million phase change memory (PCM) devices, successfully finding temporal correlations in unknown data streams. “This is an important step forward in our research of the physics of AI, which explores new hardware materials, devices and architectures,” says Evangelos Eleftheriou, PhD, an IBM Fellow and co-author of an open-access paper in the peer-reviewed journal Nature Communications. “As the CMOS scaling laws break down because of technological limits, a radical departure from the processor-memory dichotomy is needed to circumvent the limitations of today’s computers.” “Memory has so far been viewed as a place where we merely store information, said Abu Sebastian, PhD. exploratory memory and cognitive technologies scientist, IBM Research and lead author of the paper. But in this work, we conclusively show how we can exploit the physics of these memory devices to also perform a rather high-level computational primitive. The result of the computation is also stored in the memory devices, and in this sense the concept is loosely inspired by how the brain computes.” Sebastian also leads a European Research Council funded project on this topic.
Building on the devastating Mirai botnet that took major sites offline a year ago, Reaper has some scary new tricks. The Mirai botnet, a collection of hijacked gadgets whose cyberattack made much of the internet inaccessible in parts of the US and beyond a year ago, previewed a dreary future of zombie connected-device armies run amuck. But in some ways, Mirai was relatively simple—especially compared to a new botnet that's brewing. While Mirai caused widespread outages, it impacted IP cameras and internet routers by simply exploiting their weak or default passwords. The latest botnet threat, known as alternately as IoT Troop or Reaper, has evolved that strategy, using actual software-hacking techniques to break into devices instead. It's the difference between checking for open doors and actively picking locks—and it’s already enveloped devices on a million networks and counting. Now, researchers at the Chinese security firm Qihoo 360 and the Israeli firm Check Point detailed the new IoT botnet, which builds on portions of Mirai’s code, but with a key difference: Instead of merely guessing the passwords of the devices it infects, it uses known security flaws in the code of those insecure machines, hacking in with an array of compromise tools and then spreading itself further. And while Reaper hasn’t been used for the kind of distributed denial of service attacks that Mirai and its successors have launched, that improved arsenal of features could potentially allow it to become even larger—and more dangerous—than Mirai ever was. “The main differentiator here is that while Mirai was only exploiting devices with default credentials, this new botnet is exploiting numerous vulnerabilities in different IoT devices. The potential here is even bigger than what Mirai had,” says Maya Horowitz, the manager of Check Point’s research team. “With this version it’s much easier to recruit into this army of devices.” The Reaper malware has pulled together a grab-bag of IoT hacking techniques that include nine attacks affecting routers from D-Link, Netgear, and Linksys, as well as internet-connected surveillance cameras, including those sold by companies like Vacron, GoAhead, and AVTech. While many of those devices have patches available, most consumers aren’t in the habit of patching their home network router, not to mention their surveillance camera systems.
Via Roger Smith
A WSU research team for the first time has developed a computer algorithm that is nearly as accurate as people are at mapping brain neural networks—a breakthrough that could speed up the image analysis that researchers use to understand brain circuitry. For more than a generation, people have been trying to improve understanding of human brain circuitry, but are challenged by its vast complexity. It is similar to having a satellite image of the earth and trying to map out 100 billion homes, all of the connecting streets and everyone's destinations, said Shuiwang Ji, associate professor in the School of Electrical Engineering and Computer Science and lead researcher on the project. Researchers, in fact, took more than a decade to fully map the circuitry of just one animal's brain—a worm that has only 302 neurons. The human brain, meanwhile, has about 100 billion neurons, and the amount of data needed to fully understand its circuitry would require 1000 exabytes of data, or the equivalent of all the data that is currently available in the world. To map neurons, researchers currently use an electron microscope to take pictures—with one image usually containing a small number of neurons. The researchers then study each neuron's shape and size as well as its thousands of connections with other nearby neurons to learn about its role in behavior or biology. "We don't know much about how brains work," said Ji. With such rudimentary understanding of our circuitry, researchers are limited in their ability to understand the causes of devastating brain diseases, such as Alzheimer's, schizophrenia, autism or Parkinson's disease, he said. Instead, they have to rely on trial and error experimentation to come up with treatments. The National Academy of Engineering has listed improving understanding of the human brain as one of its grand challenges for the 21st century. In 2013, MIT organized a competition that called on researchers to develop automated computer algorithms that could speed up image analysis, decode and understand images of brain circuitry. As part of the competition, the algorithms are compared to work that was done by a real team of neuroscientists. If computers can become as accurate as humans, they will be able to do the computations much faster and cheaper than humans, said Ji. WSU's research team developed the first computational model that was able to reach a human level of performance in accuracy. Just as a human eye takes in information and then analyzes it in multiple stages, the WSU team developed acomputational model that takes the image as its input and then processes it in a many-layered network before coming to a decision. In their algorithm, the researchers developed an artificial neural network that imitates humans' complex biological neural networks. While the WSU research team was able to approach human accuracy in the MIT challenge, they still have a lot of work to do in getting the computers to develop complete and accurate neural maps. The computers still make a large number of mistakes, and there is not yet a gold standard for comparing human and computational results, said Ji. Although it may not be realistic to expect that automated methods would completely replace human soon, improvements in computational methods will certainly lead to reduced manual proof-reading, he added.
For the first time, researchers have sent a quantum-secured message containing more than one bit of information per photon through the air above a city. The demonstration showed that it could one day be practical to use high-capacity, free-space quantum communication to create a highly secure link between ground-based networks and satellites, a requirement for creating a global quantum encryption network.
Quantum encryption uses photons to encode information in the form of quantum bits. In its simplest form, known as 2D encryption, each photon encodes one bit: either a one or a zero. Scientists have shown that a single photon can encode even more information — a concept known as high-dimensional quantum encryption — but until now this has never been demonstrated with free-space optical communication in real-world conditions. With eight bits necessary to encode just one letter, for example, packing more information into each photon would significantly speed up data transmission.
“Our work is the first to send messages in a secure manner using high-dimensional quantum encryption in realistic city conditions, including turbulence,” said research team lead, Ebrahim Karimi, University of Ottawa, Canada. “The secure, free-space communication scheme we demonstrated could potentially link Earth with satellites, securely connect places where it is too expensive to install fiber, or be used for encrypted communication with a moving object, such as an airplane.”
In treating brain cancer, time is of the essence. A new study, in which IBM Watson took just 10 minutes to analyze a brain-cancer patient’s genome and suggest a treatment plan, demonstrates the potential of artificially intelligent medicine to improve patient care. But although human experts took 160 hours to make a comparable plan, the study’s results weren’t a total victory of machine over humans. The patient in question was a 76-year-old man who went to his doctor complaining of a headache and difficulty walking. A brain scan revealed a nasty glioblastoma tumor, which surgeons quickly operated on; the man then got three weeks of radiation therapy and started on a long course of chemotherapy. Despite the best care, he was dead within a year. While both Watson and the doctors analyzed the patient’s genome to suggest a treatment plan, by the time tissue samples from his surgery had been sequenced the patient had declined too far. IBM has been outfitting Watson, its “cognitive computing” platform, to tackle multiple challenges in health care, including an effort to speed up drug discovery and several ways to help doctors with patient care. In this study, a collaboration with the New York Genome Center (NYGC), researchers employed a beta version of IBM Watson for Genomics. IBM Watson’s key feature is its natural-language-processing abilities. This means Watson for Genomics can go through the 23 million journal articles currently in the medical literature, government listings of clinical trials, and other existing data sources without requiring someone to reformat the information and make it digestible. Other Watson initiatives have also given the system access to patients’ electronic health records, but those records weren’t included in this study. Laxmi Parida, who leads the Watson for Genomics science team, explains that most cancer patients don’t have their entire genome (consisting of 3 billion units of DNA) scanned for mutations. Instead they typically do a “panel” test that looks only at a subset of genes that are known to play a role in cancer. The new study, published in the journal Neurology Genetics, used the 76-year-old man’s case to answer two questions. First, the researchers wanted to know if scanning a patient’s whole genome, which is more expensive and time consuming than running a panel, provides information that is truly useful to doctors devising a treatment plan. “We were trying to answer the question, Is more really more?” says Parida. The answer to that question was a resounding yes. Both the NYGC clinicians and Watson identified mutations in genes that weren’t checked in the panel test but which nonetheless suggested potentially beneficial drugs and clinical trials.
Computer characters and eventually robots could learn complex motor skills like walking and running through trial and error, thanks to a milestone algorithm. "We're creating physically-simulated humans that learn to move with skill and agility through their surroundings," said Michiel van de Panne, a UBC computer science professor who is presenting this research today at SIGGRAPH 2017, the world's largest computer graphics and interactive techniques conference. "We're teaching computer characters to learn to respond to their environment without having to hand-code the required strategies, such as how to maintain balance or plan a path through moving obstacles. Instead, these behaviors can be learned." The work, called DeepLoco, offers an alternative way to animate human movement in games and film instead of the current method which makes use of actors and motion capture cameras or animators. DeepLoco allows characters to automatically move in ways that are both realistic and attentive to their surroundings and goals. In the future, two or four-legged robots could learn to navigate through their environment without needing to hand-code the appropriate rules. Using his algorithm, simulated characters have learned to walk along a narrow path without falling off, to avoid running into people or other moving obstacles, and even to dribble a soccer ball towards a goal. The method makes advanced use of deep reinforcement learning, a type of machine learning algorithm in which experience is gained through trial and error and is informed by rewards. Over time, the system progressively identifies better actions to take in given situations. "It's like learning a new sport," said van de Panne. "Until you try it, you don't know what you need to pay attention to. If you're learning to snowboard, you may not know that you need to distribute your weight in a particular way between your toes and heels. These are strategies that are best learned, as they are very difficult to code or design in any other way." The motion of humans and animals is governed not just by physics but also control. While humans learn motor control through trial and error, van de Panne says it's hard to tell how much the algorithm mimics the human learning process. After all, the computer program still learns much more slowly than a human. He began working on this type of motor learning problem when he had children; they are now 17 and 20. "I distinctly remember wondering who will learn agile walking and running skills first: my son, daughter or the algorithm?" he said. "My son and daughter beat me by a long shot." Watch a video here: https://youtu.be/15IyqCwTV4A
A future graphene-based transistor using spintronics could lead to tinier computers that are a thousand times faster and use a hundredth of the power of silicon-based computers. The radical transistor concept, created by a team of researchers at Northwestern University, The University of Texas at Dallas, University of Illinois at Urbana-Champaign, and University of Central Florida, is explained this month in an open-access paper in the journal Nature Communications. Transistors act as on and off switches. A series of transistors in different arrangements act as logic gates, allowing microprocessors to solve complex arithmetic and logic problems. But the speed of computer microprocessors that rely on silicon transistors has been relatively stagnant since around 2005, with clock speeds mostly in the 3 to 4 gigahertz range. The researchers discovered that by applying a magnetic field to a graphene ribbon (created by unzipping a carbon nanotube), they could change the resistance of current flowing through the ribbon. The magnetic field — controlled by increasing or decreasing the current through adjacent carbon nanotubes — increased or decreased the flow of current. A cascading series of graphene transistor-based logic circuits could produce a massive jump, with clock speeds approaching the terahertz range — a thousand times faster.* They would also be smaller and substantially more efficient, allowing device-makers to shrink technology and squeeze in more functionality, according to Ryan M. Gelfand, an assistant professor in The College of Optics & Photonics at the University of Central Florida. The researchers hope to inspire the fabrication of these cascaded logic circuits to stimulate a future transformative generation of energy-efficient computing.
Beyond Trillion Particle Cosmological Simulations for the Next Era of Galaxy Surveys. Researchers from the University of Zurich have simulated the formation of our entire universe with a large supercomputer. A gigantic catalogue of about 25 billion virtual galaxies has been generated from 2 trillion digital particles. This catalogue is being used to calibrate the experiments on board the Euclid satellite, that will be launched in 2020 with the objective of investigating the nature of dark matter and dark energy. Over a period of three years, a group of astrophysicists from the University of Zurich has developed and optimised a revolutionary code to describe with unprecedented accuracy the dynamics of dark matter and the formation of large-scale structures in the universe. As Joachim Stadel, Douglas Potter and Romain Teyssier report in their recently published paper, the code (called PKDGRAV3) has been designed to use optimally the available memory and processing power of modern supercomputing architectures, such as the "Piz Daint" supercomputer of the Swiss National Computing Center (CSCS). The code was executed on this world-leading machine for only 80 hours, and generated a virtual universe of two trillion (i.e., two thousand billion or 2 x 1012) macro-particles representing the dark matter fluid, from which a catalogue of 25 billion virtual galaxies was extracted. Thanks to the high precision of their calculation, featuring a dark matter fluid evolving under its own gravity, the researchers have simulated the formation of small concentration of matter, called dark matter halos, in which we believe galaxies like the Milky Way form. The challenge of this simulation was to model galaxies as small as one tenth of the Milky Way, in a volume as large as our entire observable universe. This was the requirement set by the European Euclid mission, whose main objective is to explore the dark side of the universe.
The cyber security wars of the future will be fought by good AI bots and bad ones, with the rest of us just watching to see who wins. That’s the future according to Jason Hoffman, Ericsson’s VP of Cloud Infrastructure. It isn’t quite as dark as the world being taking over by robots or sentient beings, but it’s a very realistic possibility due to the vast complications and workloads which will soon be placed on security teams. “Ironically and unfortunately, some of the people who are becoming most advanced when it comes to artificial intelligence in the security world are the ones on the offensive,” said Hoffman. “These are the cyber criminals, and one of the only ways to combat these guys will be to escalate defences to be built around artificial intelligence.” It’s a world which pits computer against computer, where Darwinism has taken a twist. The definition of ‘fittest’ moves away from strength and into the sphere of the intellectuals. But this is the end of the story, not the beginning. At the beginning, where we are right now, there is a shift in the security paradigm. In the first instance, its due to the way infrastructure is purchased and managed. In years gone, buying and securing infrastructure was relatively simple. You bought the hardware and set up restrictions surrounding the software do define who could access sensitive areas. The introduction of cloud-computing has increased accessibility, and therefore the way in which we make our life secure. One of the most attractive principles of cloud computing is the ease to scale and consume. On the operational side, this is a game changer, but for security it becomes a much more complicated task. The security paradigm has been permanently altered, as more people are now able to access sensitive areas of the machine. If one objective is to remove the threat of malicious insiders, the task has become more complex, as the ease of consumption has multiplied the number of potential malicious insiders vastly. Restrictions have to be opened up to create the cloud business model and move towards a DevOps mindset, but this involves a much more comprehensive security and governance model to be put in place, which most organizations do not currently have. Another complication is the means the shift in how infrastructure is managed. Previously, hardware has been bought, it is secured in the warehouse and then shipped to the customer. It was secure until it had served its purpose and ultimately replaced. Hoffman highlighted that a continuous stream of software updates now mean the system is only as secure as it was the last time you checked. Every update is a potential weak link in the perimeter, which again, organizations are not prepared for. It involves a complete rethink of how supply chain assurance is managed. In both these instances, the data which needs to be managed to ensure security is far too vast for any human to consider. From Hoffman’s perspective, the only option is a machine learning algorithm, which understands what would be considered normal performance from each component, and constantly monitors for the anomalies. Here, artificial intelligence is aiding the security professionals by finding the leak and then alerting, but it won’t be long before AI is the leading player.
Harvard physicists have created a new form of matter - dubbed a time crystal - which could offer important insights into the mysterious behavior of quantum systems. Traditionally speaking, crystals - like salt, sugar or even diamonds - are simply periodic arrangements of atoms in a three-dimensional lattice. Time crystals, on the other hand, take that notion of periodically-arranged atoms and add a fourth dimension, suggesting that - under certain conditions - the atoms that some materials can exhibit periodic structure across time. Led by Professors of Physics Mikhail Lukin and Eugene Demler, a team consisting of post-doctoral fellows Renate Landig and Georg Kucsko, Junior Fellow Vedika Khemani, and Physics Department graduate students Soonwon Choi, Joonhee Choi and Hengyun Zhou built a quantum system using a small piece of diamond embedded with millions of atomic-scale impurities known as nitrogen-vacancy (NV) centers. They then used microwave pulses to "kick" the system out of equilibrium, causing the NV center's spins to flip at precisely-timed intervals - one of the key markers of a time crystal. The work is described in a paper published in Nature in March. But the creation of a time crystal isn't significant merely because it proves the previously-only-theoretical materials can exist, Lukin said, but because they offer physicists a tantalizing window into the behavior of such out-of-equilibrium systems. "There is now broad, ongoing work to understand the physics of non-equilibrium quantum systems," Lukin said. "This is an area that is of interest for many quantum technologies, because a quantum computer is basically a quantum system that's far away from equilibrium. It's very much at the frontier of research...and we are really just scratching the surface." But while understanding such non-equlibrium systems could help lead researchers down the path to quantum computing, the technology behind time crystals may also have more near-term applications as well.
Neuromorphic chips are being designed to specifically mimic the human brain – and they could soon replace CPUs. Canadian startup Applied Brain Research is one of a wave of companies developing neuromorphic chips – which have several advantages over traditional CPUs. “Things like vision systems, speech systems, motion control, and adaptive robotic controllers have already been built with Nengo,” Peter Suma, a trained computer scientist and the other CEO of Applied Brain Research, tells me. Perhaps the most impressive system built using the compiler is Spaun, a project that in 2012 earned international praise for being the most complex brain model ever simulated on a computer. Spaun demonstrated that computers could be made to interact fluidly with the environment, and perform human-like cognitive tasks like recognizing images and controlling a robot arm that writes down what it’s sees. The machine wasn’t perfect, but it was a stunning demonstration that computers could one day blur the line between human and machine cognition. Recently, by using neuromorphics, most of Spaun has been run 9000x faster, using less energy than it would on conventional CPUs – and by the end of 2017, all of Spaun will be running on Neuromorphic hardware. Eliasmith won NSERC’s John C. Polyani award for that project — Canada’s highest recognition for a breakthrough scientific achievement – and once Suma came across the research, the pair joined forces to commercialize these tools. “While Spaun shows us a way towards one day building fluidly intelligent reasoning systems, in the nearer term neuromorphics will enable many types of context aware AIs,” says Suma. Suma points out that while today’s AIs like Siri remain offline until explicitly called into action, we’ll soon have artificial agents that are ‘always on’ and ever-present in our lives. “Imagine a SIRI that listens and sees all of your conversations and interactions. You’ll be able to ask it for things like - "Who did I have that conversation about doing the launch for our new product in Tokyo?" or "What was that idea for my wife's birthday gift that Melissa suggested?,” he says. “The most fundamental difference between most available AI systems of today and the biological intelligent systems we are used to, is the fact that the latter always operate in real-time. Bodies and brains are built to work with the physics of the world,” he says. Already, major efforts across the IT industry are heating up to get their AI services into the hands of users. Companies like Apple, Facebook, Amazon, and even Samsung, are developing conversational assistants they hope will one day become digital helpers. With the rise of neuromorphics, and tools like Nengo, we could soon have AI’s capable of exhibiting a stunning level of natural intelligence – right on our phones.
Via Scott Turner
The internet is much more than just the publicly available, Google-able web services most online users frequent – and that's good for free expression. Companies frequently create private networks to enable employees to use secure corporate servers, for example. And free software allows individuals to create what are called "peer-to-peer" networks, connecting directly from one machine to another. Unable to be indexed by current search engines, and therefore less visible to the general public, subnetworks like these are often called "darknets," or collective as the singular "darknet." These networks typically use software, such as Tor, that anonymizes the machines connecting to them, and encrypts the data traveling through their connections. Some of what's on the darknet is alarming. A 2015 story from Fox News reads: "Perusing the darknet offers a jarring jaunt through jaw-dropping depravity: Galleries of child pornography, videos of humans having sex with animals, offers for sale of illegal drugs, weapons, stolen credit card numbers and fake identifications for sale. Even human organs reportedly from Chinese execution victims are up for sale on the darknet." But that's not the whole story – nor the whole content and context of the darknet. Portraying the darknet as primarily, or even solely, for criminals ignores the societal forces that push people toward these anonymous networks. Our research into the content and activity of one major darknet, called Freenet, indicates that darknets should be understood not as a crime-ridden "Wild West," but rather as "wilderness," spaces that by design are meant to remain unsullied by the civilizing institutions – law enforcement, governments and corporations – that have come to dominate the internet. There is definitely illegal activity on the darknet, as there is on the open internet. However, many of the people using the darknet have a diverse range of motives and activities, linked by a common desire to reclaim what they see as major benefits of technology: privacy and free speech.
|
Currently, an interdisciplinary team of researchers are building the first-ever end-to-end simulation code to precisely capture the geology and physics of regional earthquakes, and how the shaking impacts buildings. This work is part of the DOE’s Exascale Computing Project, which aims to maximize the benefits of exascale—future supercomputers that will be 50 times faster than our nation’s most powerful system today—for U.S. economic competitiveness, national security and scientific discovery. “Due to computing limitations, current geophysics simulations at the regional level typically resolve ground motions at 1-2 hertz (vibrations per second). Ultimately, we’d like to have motion estimates on the order of 5-10 hertz to accurately capture the dynamic response for a wide range of infrastructure,” says David McCallen, who leads an ECP-supported effort called High Performance, Multidisciplinary Simulations for Regional Scale Seismic Hazard and Risk Assessments. He’s also a guest scientist in Berkeley Lab’s Earth and Environmental Sciences Area. One of the most important variables that affect earthquake damage to buildings is seismic wave frequency, or the rate at which an earthquake wave repeats each second. Buildings and structures respond differently to certain frequencies. Large structures like skyscrapers, bridges, and highway overpasses are sensitive to low frequency shaking, whereas smaller structures like homes are more likely to be damaged by high frequency shaking, which ranges from 2 to 10 hertz and above. McCallen notes that simulations of high frequency earthquakes are more computationally demanding and will require exascale computers.
The technology behind Bitcoin could touch every transaction you ever make Bitcoin was hatched as an act of defiance. Unleashed in the wake of the Great Recession, the cryptocurrency was touted by its early champions as an antidote to the inequities and corruption of the traditional financial system. They cherished the belief that as this parallel currency took off, it would compete with and ultimately dismantle the institutions that had brought about the crisis. Bitcoin’s unofficial catchphrase, “In cryptography we trust,” left no doubt about who was to blame: It was the middlemen, the bankers, the “trusted” third parties who actually couldn’t be trusted. These humans simply got in the way of other humans, skimming profits and complicating transactions. Bitcoin sought to replace the services provided by these intermediaries with cryptography and code. When you use a check to pay your mortgage, a series of agreements occur in the background between your financial institution and others, enabling money to go from your account to someone else’s. Your bank can vouch that your money is good because it keeps records indicating where every penny in your account came from, and when. Bitcoin and other cryptocurrencies replace those background agreements and transactions with software—specifically, a distributed and secure database called a blockchain. The process with which the ownership of a Bitcoin token will pass from one person to another—wherever they are, no matter what government they live under—is entrusted to a bunch of computers. Now, eight years after the first blockchain was built, people are trying to apply it to procedures and processes beyond merely the moving of money with varying degrees of success. In effect, they’re asking, What other agreements can a blockchain automate? What other middlemen can blockchain technology retire? Can a blockchain find people offering rides, link them up with people who are trying to go somewhere, and give the two parties a transparent platform for payment? Can a blockchain act as a repository and a replay platform for TV shows, movies, and other digital media while keeping track of royalties and paying content creators? Can a blockchain check the status of airline flights and pay travelers a previously agreed upon amount if their planes don’t take off on time? If so, then blockchain technology could get rid of Uber, Netflix, and every flight-insurance provider on the market. If the blockchain were a religion, Satoshi would be God. This anonymous hacker is responsible for writing the Bitcoin white paper, releasing the first Bitcoin code, and inspiring legions of blockchain developers. Many have sought to reveal his/her/their identity, but to this day that information remains secret. Those three proposed applications aren’t hypothetical—they’re just a few of the things now being built on Ethereum, a blockchain platform that remotely executes software on a distributed computer system called the Ethereum Virtual Machine. In the blockchain universe, Ethereum, which has its own cryptocurrency, called ethers, is by far the project that is most open to experimentation. But zoom out and a diverse collection of potentially disruptive innovators floods into view. New groups are pitching blockchain schemes almost daily. And the tech world’s titans don’t plan to miss out: Microsoft is offering its customers tools to experiment with blockchain applications on its Azure cloud. IBM, Intel, and others are collaborating on an open-source blockchain initiative called Hyperledger, which aims to provide the bones for business-oriented blockchains. Meanwhile, many of the largest banks—the very institutions that blockchain pioneers were trying to neutralize—have cobbled together their own version of the technology in an attempt to stay ahead of the curve. And even Bitcoin, which runs on the first and most successful blockchain, is being retrofitted for applications its designers never dreamed of.
A team of researchers with the Chinese Academy of Sciences in Beijing has announced that it has broken a record set just last month by a team at the University of Zurich in Switzerland. Since the advent of computers, space scientists have attempted to use them to create a virtual, or simulated universe. The idea is that if the universe can be simulated, it can be better studied because it can be observed over manipulated time from its birth to today. The problem with simulating the universe, of course, is that it is made up of so much stuff. Such simulations are known in the field as N-body simulations, because they become more intense as more particles are added. During the early years of such efforts, in the 1970s, computers could only handle on the order of a thousand particles. That number increased to the trillions over the past few years as computers have grown ever more powerful. The researchers with this new effort used the most powerful computer available in the world today, the Sunway TaihuLight supercomputer, built by China and based in Wuxi. Officials with the project told the press that the team had simulated the universe from birth to early expansions (approximately ten million years after the Big Bang) over the course of one hour and that the effort involved 10 trillion digital particles. The result was a virtual universe five times as big as the one created by the team in Switzerland just last month. The supercomputer gets its speed by utilizing millions of CPU cores—each able to carry out instructions independently. During the creation of the virtual universe, the team was able to make use of 10 million cores, which they described "as lots of calculations." The researchers also reported that because of the unique architecture of the Sunway TaihuLight, the team had to write almost all of the software for the project from scratch—a very labor-intensive task. They also noted that the supercomputer encountered no problems and that the simulation was terminated after just an hour because another team had booked time on the computer.
Last year, artificial intelligence (AI) research company DeepMind shared details on WaveNet, a deep neural network used to synthesize realistic human speech. Now, an improved version of the technology is being rolled out for use with Google Assistant. A system for speech synthesis — otherwise known as text-to-speech (TTS) — typically utilizes one of two techniques. Concatenative TTS involves the piecing together of chunks of recordings from a voice actor. The drawback of this method is that audio libraries must be replaced whenever upgrades or changes are made. The other technique, parametric TTS, uses a set of parameters to produce computer-generated speech, but this speech can sometimes sound unnatural and robotic. WaveNet, on the other hand, produces waveforms from scratch based on a system developed using a convolutional neural network. To begin, a large number of speech samples were used to train the platform to synthesize voices, taking into account which waveforms sounded realistic and which did not. This gave the speech synthesizer the ability to produce natural intonation, even including details like lip smacks. Depending on the samples fed into the system, it would develop a unique “accent,” which means it could be used to create any number of distinct voices if fed different data sets.
Via Nik Peachey
While Google Street View and Google Earth already give us a good view of whichever place we want to take a look at almost anywhere in the world, there is also nothing like seeing those places through the lens of another person who is actually there or who have been there. Google Earth now includes a global map of crowdsourced photos which you can consult when making travel plans, doing research for school, or just dreaming about another place far away from your home.
You’d think computers spend most of their time and energy doing, well, computation. But that’s not the case: about 90 percent of a computer’s execution time and electrical energy is spent transferring data between the processor and the memory banks, says Subhasish Mitra, a computer scientist at Stanford University. Even if Moore’s law continued on indefinitely, computers would still be limited by this memory bottleneck. Recently, in the journal Nature, Mitra and collaborators describe a new computer architecture they say addresses this problem—and that Mitra believes will improve both the energy efficiency and speed of computers by a factor of 1000. The new 3D architecture is based on novel devices including 2 million carbon nanotube transistors and over 1 million resistive RAM cells, all built on top of a layer of silicon using existing fabrication methods and connected by densely packed metal wiring between the layers. As a demonstration, the team built an electronic nose that can sense and identify several common vapors including lemon juice, rubbing alcohol, vodka, wine, and beer. These novel nanodevices are interesting in themselves, says Mitra, but computing’s problems will not be solved by switching out existing devices with new ones that are slightly better. He says the important thing about the combination of technologies in their prototype is that it enabled them to develop a new, more efficient architecture that would not be possible to make in traditional CMOS. Stacking circuits is a way to bring memory and processing closer together, but even 3D chips have a significant memory bottleneck. They are typically limited by the number and quality of connections between levels. It’s not possible to build conventional metal interconnects on top of one layer and then add another level of memory or processing because of the temperatures required—in excess of 1000 degrees Celsius to make silicon devices. Typically the layers are made separately, then bonded together and connected with relatively large, sparsely distributed connectors called through-silicon vias tens of micrometers apart, says Max Shulaker, a computer scientist at MIT. Shulaker worked with Mitra and Stanford electrical engineer H.-S. Philip Wong on the 3D nanosystem.
A study by MIT researchers shows that collections of ultracold molecules can retain the information stored in them for hundreds of times longer than previously achieved in these materials. These clusters might thus serve as “qubits,” the basic building blocks of quantum computers. Researchers have taken an important step toward the long-sought goal of a quantum computer, which in theory should be capable of vastly faster computations than conventional computers, for certain kinds of problems. The new work shows that collections of ultracold molecules can retain the information stored in them, for hundreds of times longer than researchers have previously achieved in these materials. These two-atom molecules are made of sodium and potassium and were cooled to temperatures just a few ten-millionths of a degree above absolute zero (measured in hundreds of nanokelvins, or nK). The results are described in a report this week in Science, by Martin Zwierlein, an MIT professor of physics and a principal investigator in MIT's Research Laboratory of Electronics; Jee Woo Park, a former MIT graduate student; Sebastian Will, a former research scientist at MIT and now an assistant professor at Columbia University, and two others, all at the MIT-Harvard Center for Ultracold Atoms. Many different approaches are being studied as possible ways of creating qubits, the basic building blocks of long-theorized but not yet fully realized quantum computers. Researchers have tried using superconducting materials, ions held in ion traps, or individual neutral atoms, as well as molecules of varying complexity. The new approach uses a cluster of very simple molecules made of just two atoms. “Molecules have more ‘handles’ than atoms,” Zwierlein says, meaning more ways to interact with each other and with outside influences. “They can vibrate, they can rotate, and in fact they can strongly interact with each other, which atoms have a hard time doing. Typically, atoms have to really meet each other, be on top of each other almost, before they see that there's another atom there to interact with, whereas molecules can see each other” over relatively long ranges. “In order to make these qubits talk to each other and perform calculations, using molecules is a much better idea than using atoms,” he says. Using this kind of two-atom molecules for quantum information processing “had been suggested some time ago,” says Park, “and this work demonstrates the first experimental step toward realizing this new platform, which is that quantum information can be stored in dipolar molecules for extended times.” “The most amazing thing is that [these] molecules are a system which may allow realizing both storage and processing of quantum information, using the very same physical system,” Will says. “That is actually a pretty rare feature that is not typical at all among the qubit systems that are mostly considered today.”
China now has more supercomputers among the world’s top 500 fastest machines than any other nation. Housed in the National Supercomputing Center in Wuxi, China, TaihuLight is capable of performing 93 quadrillion calculations per second. That makes it three times faster than the previous number one, Tianhe-2—also from China—which achieves speeds of 33 quadrillion calculations per second. TaihuLight uses 41,000 chips, each with 260 processor cores, to give it a grand total of 10.65 million cores. There’s also 1.3 petabytes of RAM—slightly less than Tianhe-2. The new computing champ is more efficient than its predecessor, too, drawing 15.3 megawatts of power, compared to 17.8 megawatts. Today’s news also delivers two blows to American supercomputer dominance. Notably, it's the first time that more of the world’s top 500 supercomputers are Chinese than American. China now boasts 167 systems in the rankings, while the U.S. has 165. No other country rivals either nation: third on the list is Japan, with 29 systems. Even more striking is the fact that TaihuLight doesn’t rely on Western hardware. In the past, the world’s fastest supercomputers have been built using U.S.-designed chips, with hardware from Intel or International Business Machines, or else silicon manufactured under license from Sun Microsystems. TaihuLight, however, uses Chinese-made processors. In fact, the supercomputer was funded by China’s 863 program, sometimes known as the State High-Tech Development Plan. The program was set up in 1986 by the government of the People's Republic of China, with the aim of growing the nation’s domestic technology sectors and making China independent of other countries for its hardware. To date, China’s wider chip business has lagged behind that of the U.S. and far, far behind those of Japan and South Korea. But in March the country announced it was investing $24 billion into its home-grown semiconductor industry. The state-owned chip maker XMC is already building facilities that will manufacture NAND flash-memory and DRAM chips, the first of which is expected to be operational in 2017.
A new field of physics seeking such advancements is called valleytronics, which exploits the electron's "valley degree of freedom" for data storage and logic applications. Simply put, valleys are maxima and minima of electron energies in a crystalline solid. A method to control electrons in different valleys could yield new, super-efficient computer chips. A University at Buffalo team, led by Hao Zeng, PhD, professor in the Department of Physics, worked with scientists around the world to discover a new way to split the energy levels between the valleys in a two-dimensional semiconductor. The work is described in a study published online today (May 1, 2017) in the journal Nature Nanotechnology. The key to Zeng's discovery is the use of a ferromagnetic compound to pull the valleys apart and keep them at different energy levels. This leads to an increase in the separation of valley energies by a factor of 10 more than the one obtained by applying an external magnetic field. "Normally there are two valleys in these atomically thin semiconductors with exactly the same energy. These are called 'degenerate energy levels' in quantum mechanics terms. This limits our ability to control individual valleys. An external magnetic field can be used to break this degeneracy. However, the splitting is so small that you would have to go to the National High Magnetic Field Laboratories to measure a sizable energy difference. Our new approach makes the valleys more accessible and easier to control, and this could allow valleys to be useful for future information storage and processing," Zeng said. The simplest way to understand how valleys could be used in processing data may be to think of two valleys side by side. When one valley is occupied by electrons, the switch is "on." When the other valley is occupied, the switch is "off." Zeng's work shows that the valleys can be positioned in such a way that a device can be turned "on" and "off," with a tiny amount of electricity.
Via Mariaschnee
If you have solar panels that produce more energy than you need, you can sell the excess to a utility company. But what if you could sell it to your neighbor instead? A company called LO3 Energy has developed a system that lets people buy and sell locally generated solar energy within their communities. The system uses blockchain—the electronic ledger technology that underpins the digital currency Bitcoin—to facilitate and record the transactions. Distributing energy this way is more efficient than transmitting energy over distances, said LO3’s founder, Lawrence Orsini, and would make neighborhoods more resilient to power outages, as well as helping meet demand when energy needs exceed expectations. It’s also in line with growing public support for renewable energy, distributed and decentralized energy systems, and “buy local” programs in general. At Business of Blockchain, a conference organized by MIT Technology Review and the MIT Media Lab, Orsini said that 69 percent of consumers told the technology consultancy Accenture that they were interested in having an energy-trading marketplace, and 47 percent said they planned to sign up for community solar projects. LO3 Energy launched its peer-to-peer energy transactions system, which it calls the Brooklyn Microgrid, about a year ago. The miniature utility grid connects people who have solar panels on their roofs in several parts of Brooklyn with neighbors who want to buy locally generated green energy. Like other microgrids it operates alongside, but separate from, the traditional energy grid. Blockchain makes the Brooklyn Microgrid possible, Orsini said. Participants install smart meters equipped with the technology, which track the energy they generate and consume. Records of the automatic “smart contracts” that enable neighbor-to-neighbor transactions are also tracked using blockchain. LO3 Energy hired the software maker ConsenSys to build the system, which is based on the blockchain-based distributed computing platform Ethereum. “Blockchain is a really good communications protocol for what we want to do,” Orsini said at the conference. “This isn’t just about settling energy bills,” he added. “It’s about self-organizing at the grid edge, which can’t be done with normal databases.” Could microgrids like this shake up the energy industry? At the moment, Brooklyn Microgrid consists of only 50 physical nodes, but Orsini signed a partnership with German conglomerate Siemens in November and is talking to regulators in the U.S., Australia, and Europe about expansion. He is also willing to collaborate with utilities. “We’re not putting the utilities out of business, but we want their business model to evolve,” he said.
Researchers from The University of Manchester have shown that it is possible to build a new super-fast form of computer that “grows as it computes”.
Professor Ross D King and his team have demonstrated for the first time the feasibility of engineering a nondeterministic universal Turing machine (NUTM), and their research is to be published in the prestigious Journal of the Royal Society Interface.
The theoretical properties of such a computing machine, including its exponential boost in speed over electronic and quantum computers, have been well understood for many years – but the Manchester breakthrough demonstrates that it is actually possible to physically create a NUTM using DNA molecules.
“Imagine a computer is searching a maze and comes to a choice point, one path leading left, the other right,” explained Professor King, from Manchester’s School of Computer Science. “Electronic computers need to choose which path to follow first.
“But our new computer doesn’t need to choose, for it can replicate itself and follow both paths at the same time, thus finding the answer faster.
“This ‘magical’ property is possible because the computer’s processors are made of DNA rather than silicon chips. All electronic computers have a fixed number of chips.
Via Integrated DNA Technologies
The design of 2D or 3D geometries is involved in one way or another in most of science, art and engineering activities. Modern design tools are powerful and boost the productivity of designers, but they require a lot of training, effort and time to achieve a good understanding and an efficient exploitation. LAI4D is an R+D project whose aim is to develop an artificial intelligence able to understand the user's ideas regarding spatial imagination. This technology will help improve the communication between designers and design tools by emulating human cognitive abilities for interpreting graphic representations of geometric ideas and other capabilities. The implementation has been conceived as a dual web application that can work as a 3D rendering widget or as a free light 3D CAD tool providing the adequate environment for the project. This CAD tool incorporates a special design assistant in charge of extracting conceptual geometries from pictures or sketches provided by the user as input. Additionally LAI4D tries to reduce the inherent complexity of professional design tools which, despite being suitable for experienced users, are almost unreachable for other people not trained in the usage of CAD systems and only in need of an occasional use. The selected implementation approach not only allows the users an easy access to the tool, but is also an excellent mean to build a community of designers that will provide the necessary feedback for the system in order to make it bigger and smarter. Follow this link to try the LAI4D designer. Beginner's tutorial: This tutorial is the recommended introduction for all persons new to LAI4D. It is intended to teach the basics of design in 20 minutes. It shows step by step how to create a simple 3D geometry from the idea up to the publishing of the drawing on the Internet using the easiest tools. Thanks to this exercise the user will understand the working philosophy of LAI4D, will be able to generate polyhedral surfaces and polygonal lines with colors, and will learn to share designs through the free online sharing service of LAI4D. The created geometry can be inspected in the link: cubicle_sample.
|
