ACM TechNews

RSS

Timely Topics for IT Professionals

ACM TechNews Home

Recent News

Google's Driverless Cars Run Into Problem: Cars With Drivers

The New York Times

Researchers at Google and other companies pursuing self-driving cars are challenged by the fact that automated vehicles, which are programmed to obey the letter of the law and traffic safety rules, may have trouble fitting in with autos driven by people, who do not always adhere to such rules. "The real problem is the car is too safe," says Donald Norman, director of the Design Lab at the University of California, San Diego. "[Driverless cars] have to learn to be aggressive in the right amount, and the right amount depends on the culture." Google cars routinely make quick, evasive maneuvers or practice caution in ways that are out of alignment with other vehicles on the road. Following the most cautious approach has led to 16 crashes involving Google cars in the last six years, with Google blaming human error for every collision. Although the wide use of self-driving cars may alleviate these concerns, researchers say the near-term problem is finding ways to enable humans and machines to work together safely. Google's Courtney Hohne says researchers now are testing ways of "smoothing out" the relationship between humans and the car's software. Dmitri Dolgov, head of software for Google's Self-Driving Car Project, says he has learned the initiative demonstrates the need for human drivers to be "less idiotic."

From "Google's Driverless Cars Run Into Problem: Cars With Drivers"
The New York Times (09/01/15) Matt Richtel; Conor Dougherty
View Full Article - May Require Free Registration

"Molecular Tweeting" Could Hold the Key to Busting Superbugs

Scientific American

Carnegie Mellon University (CMU) researchers are using the metaphor of Twitter to help them better understand the ways communication among bacteria can lead to antibiotic resistance. Bacteria communicate by exchanging signaling molecules in a process known as quorum sensing. This process contributes to the ways groups of bacteria coordinate their efforts to create biofilms that can protect them from antibiotics. A team lead by CMU professor Radu Marculescu has developed a network model that uses the metaphor of Twitter to explain these behaviors. The researchers created their simulations by employing a trio of computer models of bacterial behavior. In the first, bacteria act like active Twitter users by sending out and passing along molecular messages, tweeting and retweeting them. In the second group, bacteria send out their own signals but do not pass along those of other bacteria: they tweet, but do not retweet. In the final model, the bacteria do not send any messages. The researchers think their simulations can help illuminate how antibiotic resistance arises and could even be used to create personalized treatment plans that will have the greatest possible efficacy without inducing antibiotic resistance. The researchers will present their findings this month at the ACM Conference on Bioinformatics, Computational Biology, and Health Informatics in Atlanta, and at ACM's second International Conference on Nanoscale Computing and Communication in Boston.

From ""Molecular Tweeting" Could Hold the Key to Busting Superbugs"
Scientific American (08/31/15) Larry Greenemeier
View Full Article

Tired of Memorizing Passwords? A Turing Award Winner Came Up With This Algorithmic Trick

IDG News Service

Carnegie Mellon University professor Manuel Blum, who received the ACM A.M. Turing Award in 1995, believes online passwords can become more manageable and secure via an algorithmic method to create what he calls "human computable" passwords.  The technique has people learn an algorithm and personal key beforehand, and then use them with the website's name to generate and regenerate their own unique passwords on the spur of the moment for any website at any time.  Learning the key and algorithm only has to be done once, and this spares users from the burden of memorizing passwords, according to Blum.  He says the basic concept is the algorithm and key give the user an alternate letter or number for each letter in a website's name, and that transformed set of values becomes the user's site-specific password.  There are many possible algorithms that could be employed, while for sites requiring special characters, the user could get into the habit of adding a few to the algorithmic results.  Blum says the system would present difficulty to hackers, and notes "as long as you don't give away more than a few passwords, you'll be secure."

From "Tired of Memorizing Passwords? A Turing Award Winner Came Up With This Algorithmic Trick"
IDG News Service (08/31/15) Katherine Noyes
View Full Article

How Artificial Intelligence Can Fight Air Pollution in China

Technology Review

IBM is testing a prototype computer system that is capable of learning to predict the severity of air pollution in different parts of Beijing, which is surrounded by many factories fueled by coal.  The system uses large quantities of data from several different models, and could eventually offer specific recommendations on how to reduce pollution.  For example, the system could recommend closing certain factories or temporarily restricting the number of vehicles on the road.  The system can generate high-resolution forecasts 72 hours ahead of time.  "Our researchers are currently expanding the capability of the system to provide medium- and long-term [up to 10 days ahead] as well as pollutant-source tracking, 'what-if' scenario analysis, and decision support on emission reduction actions," says IBM Research China director Xiaowei Shen.  IBM's cognitive computing approach encompasses natural-language processing and statistical techniques originally developed for the Watson supercomputer.  IBM uses data provided by the Beijing Environmental Protection Bureau to refine its models, and Shen says the predictions have a resolution of a kilometer and are 30 percent more precise than those derived through conventional approaches. He also notes the system uses “adaptive machine learning” to determine the best mix of models to use.

From "How Artificial Intelligence Can Fight Air Pollution in China"
Technology Review (08/31/15) Will Knight
View Full Article

Quantum Computer that 'Computes Without Running' Sets Efficiency Record

Phys.org

Researchers at the University of Science and Technology of China (USTC) and Yale University have demonstrated a more efficient way to build a quantum computer that computes without running, a process called counterfactual computation (CFC). The researchers developed a generalized CFC system with an efficiency of 85 percent and the potential to reach 100 percent, a breakthrough they say could lead to a much greater variety of applications. "The main keys to achieving high-efficiency CFC include the utilization of exotic quantum features [quantum superposition, quantum measurement, and the quantum Zeno effect], as well as the use of a generalized CFC protocol," says USTC professor Jiangfeng Du.  By "not running," the scientists mean the computer stays in its "off" subspace for the entire computation.  Physically maintaining the computer in the "off" subspace involves controlling the spin properties of a diamond system, which acts as a quantum switch.  "By detecting its state, we get the information that is 'programmed' in the computer, although the computer has not run," says Yale professor Liang Jiang.  Previous experimental CFC protocols have faced a counterfactual efficiency limit of 50 percent, but the generalized CFC does not face this limit, which enabled the researchers in the new study to experimentally demonstrate an efficiency of 85 percent at 17 pulse repetitions.

From "Quantum Computer that 'Computes Without Running' Sets Efficiency Record"
Phys.org (08/31/15) Lisa Zyga
View Full Article

Gaming Computers Offer Huge, Untapped Energy Savings Potential

Lawrence Berkeley National Laboratory

Lawrence Berkeley National Laboratory (Berkeley Lab) researchers recently conducted an investigation of the aggregate global energy use of personal computers designed for gaming, and found gamers can achieve energy savings of more than 75 percent by changing some settings and swapping out some components, while improving reliability and performance.  The research corresponds to a potential estimated savings of $18 billion per year globally by 2020.  Gaming computers represent only 2.5 percent of the global installed personal computer base but account for 20 percent of the energy use. The researchers note there is ample opportunity for consumers, manufacturers, and policymakers to save energy.  In terms of regulation, displays and power supplies currently are the only components that have energy ratings, but additional ratings for motherboards, hard drives, peripherals, and other parts are "an opportunity area," says Berkeley Lab's Evan Mills.  The researchers built five gaming computers with progressively more efficient component configurations, then followed industry protocols for benchmarking performance while measuring energy use.  They recorded a 50-percent reduction in energy use while performance remained essentially unchanged.  The experiment achieved additional energy savings via operational settings to certain components.  "The efficient systems run cooler and quieter, both of which are desirable attributes among gamers," Mills says.

From "Gaming Computers Offer Huge, Untapped Energy Savings Potential"
Lawrence Berkeley National Laboratory (08/31/15) Jon Weiner
View Full Article

Back to School With New Cyberlearning Tools

National Science Foundation

A new cyberlearning tool created by researchers at the Concord Consortium and Northwestern University aims to increase students' comprehension and engagement in science class by providing them with scientific data gathering, analysis, and modeling capabilities. Concord CEO Chad Dorsey says today's science curriculum often fails to engage students because of the rote nature of the exercises. The new InquirySpace tool, developed by Dorsey and Northwestern University professor Uri Wilensky, combines sensors, hardware, and software to enable students to engage in more free-form investigations. One component is the Common Online Data Analysis Platform, which enables students to create scientific models and simulations using data they have gathered. Another core component is the NetLogo system developed by Wilensky, which enables students to simulate actions and reactions in large, complex systems such as the immune system, economies, and the climate. InquirySpace also includes the Concord Consortium's Molecular Workbench software, which enables students to simulate various chemical and physical phenomena. All of the InquirySpace tools are open source and freely available so teachers can modify and alter them to best fit their own classrooms.

From "Back to School With New Cyberlearning Tools"
National Science Foundation (08/27/15) Aaron Dubrow
View Full Article

Quantum 'Spookiness' Passes Toughest Test Yet

Nature

The toughest test of quantum theory ever conducted has verified the "spooky action at a distance," in which manipulating one object instantaneously seems to affect another, remote object, is inherent to the quantum domain.  The test represents a new assessment of physicist John Bell's test for differentiating between Einstein's hidden variables and the spooky interpretation of quantum mechanics.  Bell theorized concealed variables can explain correlations only up to some maximum limit, and if that level is surpassed, then Einstein's model must be incorrect.  Previous experiments were susceptible to certain loopholes, which force testers to assume the properties of the photons they capture are representative of the entire set.  Delft University of Technology researchers report closing these loopholes via entanglement swapping.  They began with two unentangled electrons in diamond crystals held in different labs on campus, 1.3 kilometers between them.  Each electron was individually entangled with a photon, and both photons were transmitted to a third location where the two photons were entangled with each other, causing entanglement in their paired electrons as well.  The researchers produced 245 entangled electron pairs and exceeded Bell's limits.  One expert says this type of test has ramifications for quantum cryptography, and could potentially make it hackable.

From "Quantum 'Spookiness' Passes Toughest Test Yet"
Nature (08/27/15) Zeeya Merali
View Full Article

Plasmonics: Close to the Point of More Efficient Chips

A*STAR Research

A*STAR researchers in Singapore have used the tip of a scanning tunneling microscope (STM) to generate electromagnetic waves known as surface plasmon polaritons in a gold grating and shown the direction of travel of these waves can be controlled. The researchers note this is a step toward the development of plasmonic chips, which use plasmons rather than electrons to transfer and process data.  They say plasmonic chips could be much faster and potentially more energy-efficient than current electronic chips.  The researchers, led by Joel Yang and Zhaogang Dong, studied controlling the traveling direction of plasmons in a gold grating both theoretically and experimentally.  During testing, the researchers moved the STM tip relative to the edge of the gold grating and observed the generated light using an inverted microscope.  "Depending on how far we drop the pebble from the barrier for each lane, we can get waves that preferentially move away from the barrier and even across lanes," Yang says.  The ability to control the direction of the waves comes from the surface plasmon polariton reflected from the grating edge interfering with the one at the STM probe.  The researchers modeled this process on a computer and found a good match with the experimental results.  They say this breakthrough could be useful in developing ways to replace wires between chips with optical connectors.

From "Plasmonics: Close to the Point of More Efficient Chips"
A*STAR Research (08/26/15)
View Full Article

I've Taught Computers to Paint Portraits--and How to Code

New Scientist

Falmouth University professor Simon Colton's area of focus is software that behaves in a manner that would be deemed creative if observed in humans. In an interview, Colton describes his work with software dubbed HR, which is programmed to make its own discoveries.  He notes one such discovery HR made was of the algebraic classification of Latin squares, while another was the independent derivation of Goldblach's conjecture.  Colton says designing software to make discoveries involves feeding it data that you want to learn something about, but instead of seeking known unknowns--as with machine learning--it attempts to find unknown unknowns.  "So we teach it how to do general things rather than specifics," Colton says.  "That contradicts most of what we do in computer science, which is to make sure software does exactly what you want."  Another program Colton developed, the Painting Fool, creates portraits, and he says mathematicians easily accept computers as creative if they produce great results repeatedly--but persuading artists is another matter altogether.  Colton speculates true computer discovery will only arrive when software becomes capable of self-programming.

From "I've Taught Computers to Paint Portraits--and How to Code"
New Scientist (08/26/15) Douglas Heaven
View Full Article

What Human Emotions Do We Really Want of Artificial Intelligence?

The Conversation

Queensland University of Technology's David Lovell proposes a new test for artificial intelligence (AI) that transcends what he perceives as shortcomings in the Turing and Lovelace tests. Dubbing it the Frampton Test, Lovell says passing it would require an AI system "to give a convincing and emotionally appropriate response to a situation that would arouse feelings in most humans." The impetus for Lovell's proposal was a panel discussion at the Robotronica 2015 event, which led to questions about the kinds of emotions people would want AIs to express. Lovell writes empathy topped the list of human qualities people wanted machines to have, "perhaps because it goes beyond mere recognition ('I see you are angry') and demands a response that demonstrates an appreciation of emotional impact." In contrast, the Turing test seeks to assess the ability of a machine to engage in a conversation with a person to the point the person is fooled into thinking the machine is human. Meanwhile, the Lovelace test emphasizes creativity, with the AI being able to compose creative works without the programmer being able to account for how it produces them.

From "What Human Emotions Do We Really Want of Artificial Intelligence?"
The Conversation (08/28/15) David Lovell
View Full Article

Reflective Satellites May Be the Future of High-End Encryption

Ars Technica

A group of Italian researchers has demonstrated the possibility of using quantum key distribution via a satellite, which in theory means any two parties with a view of a satellite can exchange keys.  The researchers compared the polarization states they detected to the pulses of light they sent, and were able to determine the newer satellites did preserve polarization, while older satellites generated more errors.  The test showed the error rate was low enough that a key could be shared via quantum states. In the new approach, the satellite would contain optics that would modify the polarization of the light at the satellite.  Interception after this point is detectable because the reflected signal is at the single photon level.  The researchers say the most important aspect of this new method is to ensure the polarization state sent to the satellite does not reveal the polarization state reflected from the satellite. The researchers achieved this by sending pulses of light that are circularly polarized because they can be filtered to two pairs of linearly polarized states at the satellite.

From "Reflective Satellites May Be the Future of High-End Encryption"
Ars Technica (08/24/15) Chris Lee
View Full Article

Is a Cambrian Explosion Coming for Robotics?

IEEE Spectrum

New technological developments have begun to accumulate that are fomenting an explosion in the diversity and application of robotics, prompting former U.S. Defense Advanced Research Projects Agency (DARPA) program manager Gill A. Pratt to suggest a "Cambrian Explosion" of robotics is coming. Pratt is referring to a period in the early history of life on Earth in which the number and variety of animal species dramatically increased over a short time. Pratt says several trends are feeding the potential for explosive growth in robotics, including exponential growth in computing performance and data storage, the growing ubiquity of wireless communications infrastructure, faster Internet speeds, the advent of three-dimensional printing, and improvements in both electrical energy storage and the energy efficiency of electronics. Pratt also points to the rise of deep-learning algorithms and the concept of cloud robotics, in which every robot learns from the experiences of every other robot, as major drivers of growth in the realm of robotics as they accelerate and simplify the process of robots acquiring new skills and capabilities. However, he says it is very difficult to predict when, or if, a Cambrian Explosion of robotics might occur, and what effects it would have on society and the economy.

From "Is a Cambrian Explosion Coming for Robotics?"
IEEE Spectrum (08/31/15) Gill A. Pratt
View Full Article