Association for Computing Machinery
Welcome to the January 23, 2017 edition of ACM TechNews, providing timely information for IT professionals three times a week.

ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).


Data Mining Solves the Mystery of Your Slow Wi-Fi Connection
Technology Review (01/20/17)

Slow or failed connectivity is a common problem for Wi-Fi users, and Changhau Pei and colleagues at Tsinghua University in China have discovered why this happens and how to avoid it by measuring the time it took for 400 million different Wi-Fi sessions to connect. The researchers compiled data from the Wi-Fi Manager app for Android, and calculated a 45-percent rate of connection failure, while 15 percent of connections take longer than five seconds. Using a data-mining algorithm, the team determined factors underlying failed links and long connection times. The most significant factor is a network's status as public or private, as private networks are faster and have higher rates of connection success. The mobile device's operating system is another significant factor, with the researchers citing a highly customized version of Android known as FlyMe as one culprit. They also note the chipsets in both the mobile device and the access point can affect connection times, with slower chips taking longer. The test results led to the creation of an algorithm that avoids the most obvious compromises to accelerate connection times; for example, by selecting access points only on private networks. The researchers say the algorithm slashes connectivity failures rates to 3.6 percent and accelerates connection times 10-fold.

Researchers Build Carbon Nanotube Transistors That Outperform Those Made with Silicon (01/20/17) Bob Yirka

Researchers at Peking University in China say they have built a carbon nanotube-based working transistor that outperforms larger transistors made with silicon. Instead of growing carbon nanotubes that had certain desired properties, the researchers grew some and placed them randomly on a silicon surface. They then added electronics that would work with the properties they had, which allowed for the building of a carbon nanotube transistor that could be tested to see if it would verify theories about its performance. The researchers also built a new kind of electrode by etching very tiny sheets of graphene, which resulted in a very tiny transistor capable of moving more current than a standard complementary metal-oxide semiconductor transistor using only half of the normal voltage amount. In addition, the new transistor was faster due to a much shorter switch delay, thanks to a gate capacitance of only 70 femtoseconds. The researchers say their work offers physical evidence that money being spent on research into carbon nanotubes as a viable replacement for silicon will pay off if a way to mass-produce them can be found.

Faster Websites With Fewer Bugs
MIT News (01/20/17) Larry Hardesty

Researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory last week presented a new system that automatically manages database query caching for Web applications written in Ur/Web at the ACM SIGPLAN Symposium on Principles of Programming Languages (POPL 2017) in France. Ur/Web enables developers to fully specify their sites' functionality using only one programming language. The Ur/Web compiler then automatically generates all the different types of code needed to power a site while supporting certain performance and security guarantees. Because the system is a modification of the compiler, Ur/Web users only have to recompile their existing code to get all of the advantages of database caching, while the language does not change. The compiler first analyzes the Ur/Web code and determines what data to cache and how to organize it. Then it sifts through the code, comparing every operation that updates a value in the database with every function that queries the database, ascertaining which cached values need to be invalidated when, and adding suitable cache-invalidation commands in the appropriate places. Experiments with two websites built using Ur/Web found the new system's automatic caching delivered twofold and 30-fold speedups.

Algorithm Mimics Evolution to Aid Disaster Recovery
Government Computer News (01/19/17) Patrick Marshall

Lehigh University researchers have developed the Algorithm with Multiple-Input Genetic Operators (AMIGO), a new disaster-management tool that quickly weighs all the available information about infrastructure, the damage sustained, and the repair resources available. AMIGO then presents several optimal recovery strategies to a decision maker who can choose from the best alternatives. By weighing only the options that fit within the parameters of a city's recovery goals, the "genetic algorithms mimic the process of biological evolution to find the optimal solutions of a problem," says Lehigh University professor Paolo Bocchini. The researchers tested the effectiveness of AMIGO by analyzing a hypothetical earthquake in the San Diego, CA, region using information from the National Bridge Inventory, a database maintained by the Federal Highway Administration. The researchers identified the 80 bridges in the area that would have been the most damaged, and applied AMIGO to determine the best restoration strategy. AMIGO is part of the larger, U.S. National Science Foundation-funded Probabilistic Resilience Assessment of Interdependent Systems (PRAISys) project, which is aimed at modeling the interdependencies of critical infrastructure systems and their recovery after disasters. Bocchini says the ability to work with different levels of input is essential because the data available on much of the critical infrastructure is inconsistent.

Making AI Systems That See the World as Humans Do
Northwestern Now (01/19/17) Amanda Morris

Researchers at Northwestern University have devised a new computational model that executes human-level performance on a standard intelligence test. Northwestern professor Ken Forbus says the model performs in the better-than-average 75th percentile for U.S. adults, noting "the problems that are hard for people are also hard for the model, providing additional evidence that its operation is capturing some important properties of human cognition." The model uses the CogSketch artificial intelligence (AI) platform as its foundation, which is capable of solving visual problems and understanding sketches to deliver immediate, interactive feedback. CogSketch also has a computational model of analogy, based on Northwestern professor Dedre Gentner's structure-mapping theory. In collaboration with former Northwestern researcher Andrew Lovett, Forbus developed the system and tested it on Raven's Progressive Matrices, a nonverbal test for quantifying abstract reasoning. "The Raven's test is the best existing predictor of what psychologists call 'fluid intelligence,' or the general ability to think abstractly, reason, identify patterns, solve problems, and discern relationships," Lovett says. "Our results suggest that the ability to flexibly use relational representations, comparing and reinterpreting them, is important for fluid intelligence." Forbus says their work "provides an important step toward [AI] understanding visual reasoning more broadly."

Heartbeat Could Be Used as Password to Access Electronic Health Records
Binghamton University (01/18/17)

Binghamton University researchers say they have developed a new way to protect personal electronic health records using a patient's own heartbeat. Traditional encryption solutions "are gradually replacing clinic-centered healthcare, and we wanted to find a unique solution to protect sensitive personal health data with something simple, available and cost-effective," says Binghamton professor Zhanpeng Jin. The researchers encrypted patient data using a person's unique electrocardiograph (ECG) as the key to lock and unlock the files. The patient's heartbeat is essentially the password to access their electronic health records. "While ECG signals are collected for clinical diagnosis and transmitted through networks to electronic health records, we strategically reused the ECG signals for the data encryption," Jin says. "Through this strategy, the security and privacy can be enhanced while minimum cost will be added." However, an ECG can change due to age, illness, or injury, so the researchers are working on ways to incorporate those variables. The identification scheme is a combination of previous work using a person's unique brainprint instead of traditional passwords for access to computers and buildings, combined with cybersecurity work from Binghamton professors Linke Guo and Yu Chen.

Microbiologists Make Big Leap in Developing 'Green' Electronics
UMass Amherst News (01/18/17) Janet Lathrop

Scientists at the University of Massachusetts Amherst (UMass Amherst) have discovered a type of microbial nanowire that could help researchers develop sustainable conducting materials for electronic devices. Unlike synthetic nanowires requiring toxic chemicals, high temperatures, and expensive metals to produce, natural nanowires can be mass-produced using inexpensive and renewable raw materials in bioreactors with much lower energy inputs. In a new study supported by the U.S. Office of Naval Research, the UMass Amherst scientists isolated the nanowire-assembling gene from one bacterium, Geobacter metallireducens, and inserted it into a second member of the species, G. sulfurreducens. The result was a bacterium that makes nanowires 5,000 times more conductive than G. sulfurreducens would produce naturally. G. metallireducens nanowires' high conductivity is attributed to its abundance of aromatic amino acids, which support better connections for electron transfer along the protein filaments. The researchers believe the new nanowire offers unprecedented potential for building conductive materials, electronic devices, and sensors for medical and environmental applications. They say as scientists discover more about the mechanisms of nanowire conductivity, eventually better wires could be constructed with specially designed genes.

U.S. Soldiers Will Be Armed With Machine Translators to Kill Communication Woes
Quartz (01/19/17) Ephrat Livni

The U.S. military has been rolling out pieces of its Machine Foreign Language Translation System (MFLTS) since 2011 and will soon reach full deployment of the platform, which runs on military and commercial devices. Translation services are crucial for soldiers in the U.S. Army stationed abroad, but interpreters are not always available. The new system is not designed to replace human translators completely, but it is meant to complement their work and provide assistance to soldiers unable to be accompanied by their own interpreters. Moreover, soldiers can use the software to check the accuracy of their human interpreter's translations. The translation system currently supports spoken Pashto and Iraqi Arabic and written Modern Standard Arabic, and the Army is considering adding Dari, a Persian language spoken in Afghanistan, and Sorani Kurdish. MFLTS comes pre-loaded with common phrases and questions that soldiers can play to communicate with locals, and soldiers rely on the machine translator to understand the reply to their automated questions. MFLTS product director Michael Beaulieu says the application fills an important communication gap. "If you can't talk to the people who you are trying to win the hearts and minds of, it is kind of hard to win a counterinsurgency," he notes.

Soft Robot Helps the Heart Beat
Harvard University (01/18/17) Leah Burrows

Researchers at Harvard University and Boston Children's Hospital have developed a customizable soft robot that twists and compresses in sync with a beating heart, augmenting cardiovascular functions compromised by heart failure. The new device does not come into contact with blood and eliminates the need for a patient to take blood thinner medications. The device's thin silicone sleeve uses soft pneumatic actuators to mimic the outer muscle layers of a human heart. The actuators twist and squeeze the sleeve in a motion similar to the way a heart beats. The device is attached to an external pump, which uses air to power the actuators. The sleeve also is customizable, so patients with more weakness on one side of the heart can receive more assistance on that side from the actuators. More research must be done before the device can be implanted in humans, but researchers believe the robot will create valuable new treatment options for people with heart failure. "Soft robotic devices are ideally suited to interact with soft tissue and give assistance that can help with augmentation of function, and potentially even healing and recovery," says former Harvard postdoctoral student Ellen T. Roche.

UC Berkeley Doctoral Student Reports Vulnerability in WhatsApp
The Daily Californian (01/17/17) Pamela Larson

A University of California, Berkeley researcher has found that WhatsApp, an end-to-end encrypted messaging platform, has a security flaw that makes it susceptible to messages being intercepted. Tobias Boelter, a second-year electrical engineering and computer science doctoral student, says the main problem is WhatsApp will not provide the option to prevent a message from being sent after a security key change. A security key can change for many reasons, including if someone is trying to intercept a message. The effort to intercept must occur before a message is received by the intended recipient, and users are not notified of a security key change until after a message is sent. To prevent interception, people could notify their contacts before doing something that would change their security key. Boelter's colleague Moxie Marlinspike recommended Signal as an alternative app for secure end-to-end encryption communication. WhatsApp co-founder Brian Acton says the vulnerability is not a backdoor, and he notes the communication platform is designed to maintain its 1 billion users' access to their messages.

Talking to Children about STEM Fields Boosts Test Scores and Career Interest
UChicago News (IL) (01/17/17) Mark Peters

Parents who talk with their high schoolers about the relevance of science and math can increase their competency and career interest in the fields, according to a new University of Chicago study. The study found a 12-percent increase in test scores on the math and science ACT exam for students whose parents were provided with information on how to convey the importance of science, technology, engineering, and math (STEM). The research provides new insights as U.S. policymakers look to increase the number of students going into STEM fields. "Parents are potentially an untapped resource for helping to improve the STEM motivation and preparation of students," says University of Chicago researcher Christopher S. Rozek. "We could move the needle by just encouraging parents to have these conversations about the relevance of math and science." The researchers focused on expectancy-value theory and, more specifically, on the concept that individuals make choices depending on the relevance or usefulness to a current or future goal. The researchers designed materials that help parents talk to their children about the relevance of STEM fields, and then tracked a variety of outcomes over several years. When parents were provided with the STEM relevance information, their children showed improved math and science ACT scores.

Crowdfunding Expands Innovation Financing to Underserved Regions
Berkeley News (01/13/17) Brett Israel

University of California, Berkeley (UC Berkeley) researchers found crowdfunding platforms open a funding stream to startups in regions that have suffered from a venture capital drought. The researchers analyzed data from 2009 to 2015 on successful crowdfunding campaigns and venture capital investments. In addition, they focused on investments that could have reasonably been funded by either crowdfunding or venture capital. The researchers identified 55,005 crowdfunding projects in categories similar to the industries in which venture capitalists invested, and 17,493 venture capital investments in industries engaged in activities similar to those of crowdfunding campaigns. They then used this dataset to map crowdfunding projects and venture capital investments by county and by year. The researchers found although the typical crowdfunding campaign involved a smaller amount of money, they covered a broader geographic area of the country. For example, cities such as Chicago, Los Angeles, and Seattle have experienced the largest number of successful crowdfunding campaigns, even though those cities have not been magnets for venture capitalists' investments. "If this phenomenon continues, crowdfunding could begin to address regional inequality in entrepreneurial financing, through both direct crowdfunding investment and induced venture capital investment," says Lee Fleming, faculty director of the Coleman Fung Institute for Engineering Leadership at UC Berkeley.

Abstract News © Copyright 2017 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: [email protected]