MIT CSAIL's Minerva Video Protocol Reduces Buffering, Pixelation


Researchers at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Laboratory have developed a video protocol that leverages video player state data and file properties to optimize congestion control. The team said the Minerva protocol significantly reduces buffering and pixelation, without changing underlying infrastructure. Via techniques and distributed algorithms that capture the relationship between bandwidth and experience quality, Minerva dynamically adapts video streaming rates for fairness (making the video experience as similar as possible for all viewers), even when lacking explicit data on competing video clients; when several clients share a bottleneck connection, their rates converge to a bandwidth allocation that does not disrupt other Internet traffic. Said MIT's Mohammad Alizadeh, “If five people in your house are all streaming video at once, [Minerva] can analyze how the various videos' visuals are affected by download speed. It then uses that information to provide each video with the best possible visual quality without degrading the experience for others."

From "MIT CSAIL's Minerva Video Protocol Reduces Buffering, Pixelation"
VentureBeat (08/18/19) Kyle Wiggers
View Full Article

2 Quakes in 2 Days, No Warning from ShakeAlertLA. Now the App Is Getting Reworked

The Los Angeles Times

An early-warning earthquake app's failure to alert Los Angeles residents of tremors from two quakes in July prompted an upgrade so weaker tremors will set off warnings. The U.S. Geological Survey's ShakeAlert network detected the quakes about 49 seconds before major tremors hit Los Angeles County on July 5, but the ShakeAlertLA app remained silent. The app's lowered threshold means computers will immediately analyze any quake of magnitude 4.5 or higher to determine whether weak shaking is expected for the county. The previous threshold analyzed quakes of magnitude 5 or greater, and would trigger an alert only if the system predicted light (or greater) shaking.

From "2 Quakes in 2 Days, No Warning from ShakeAlertLA. Now the App Is Getting Reworked"
The Los Angeles Times (08/14/19) Rong-Gong Lin
View Full Article - May Require Paid Subscription

Researchers Use Blockchain to Drive Electric Vehicle Infrastructure

University of Waterloo News

Researchers in the Cheriton School of Computer Science and the Department of Management Science of Canada’s University of Waterloo have incorporated blockchain into energy systems, which could expand charging infrastructure for electric vehicles (EVs). An open blockchain platform will give EV owners, property owners, and charging service operators access to charging data, and alert them to tampering; EV owners will be able to see whether they are being overcharged for charging their vehicles, and property owners will be alerted to instances of underpayment. Said Waterloo’s Christian Gorenflo, “Mitigating trust issues in EV charging could result in people who have charging stations and even those who just have an outdoor outlet being much more willing to team up with an EV charging service provider, resulting in much better coverage of charging stations.”

From "Researchers Use Blockchain to Drive Electric-Vehicle Infrastructure"
University of Waterloo News (08/14/19)
View Full Article

What Your Voice Reveals About You

The Wall Street Journal

Technology can detect nuances in the human voice that offer clues to a person's likely location, medical conditions, and even physical features. For example, voice-biometric and recognition software used by Nuance Communications examines factors like the pitch, rhythm, and dialect of speech, as well as vocabulary, grammar, and sentence structure, to detect the gender, age, and linguistic background of callers and whether a voice is synthetic or recorded. It helped one bank determine that a single person was responsible for tens of millions of dollars of theft. Winterlight Labs, meanwhile, parses features in speech and works with Janssen Pharmaceuticals to try to detect Alzheimer's in older patients who, for example, tend to use words they acquired earlier in life as their recent memories deteriorate.

From "What Your Voice Reveals About You"
The Wall Street Journal (08/13/19) Sarah Krouse
View Full Article - May Require Paid Subscription

SMU Researchers Find a New Way to Snoop with Smartphones. Should You Be Worried?

Dallas Morning News

Researchers at Southern Methodist University (SMU) created an app that could be used to determine what a computer user is typing. If a few smartphones equipped with the app are placed on the same table as a computer, the app can correctly guess nearly half of all the keystrokes typed on the computer, while about 25% of the words can be "perfectly translated." The app does not do a particularly good job of identifying passwords because it is trained to pick up conversations, while passwords are often a nonsensical collection of letters and numbers. However, by using the app to narrow down which characters are typed, "instead of taking years to decode someone's password, it may take days," according to SMU researcher Eric Larson.

From "SMU Researchers Find a New Way to Snoop with Smartphones. Should You Be Worried?"
Dallas Morning News (08/13/19) Jordan Wilkerson
View Full Article

ASU Researchers Use New Tools of Data Science to Capture Single Molecules in Action


Arizona State University (ASU) scientists tapped new data science and statistical modeling tools to enhance single-molecule fluorescence techniques, a breakthrough that could overcome the long acquisition times and noisy signals of fluorescence correlation spectroscopy. ASU's Steve Presse and colleagues combined data science and Bayesian nonparametrics to achieve this upgrade. Said ASU's Marcia Levitus, "Old strategies limited our ability to probe anything but slow processes, leaving a vast number of interesting biological questions involving faster chemical reactions out of reach. Now we can begin asking questions on processes resolved in short order."

From "ASU Researchers Use New Tools of Data Science to Capture Single Molecules in Action"
ASU Now (08/14/19)
View Full Article

How a State Plans to Turn Coal Country into Coding Country

The New York Times

Wyoming has mandated that all its K-12 public schools offer computer science (CS) in an effort to transition from its economic reliance on fossil-fuel industries. The state's 48 school districts must implement CS education at all grade levels by the time the 2022-2023 school year starts. The U.S. Bureau of Labor Statistics estimated that computer coding and software development comprise fewer than two jobs per 1,000 in Wyoming, versus 19 per 1,000 in the state of Washington. State CS standards issued this spring stipulate that all Wyoming students must learn what an algorithm is, understand concepts like loops, examine technology's societal impact, and become capable of writing their own code. Microsoft and Wyoming's Department of Education hope to supply CS training for at least one teacher in each school, partly through cooperation with the University of Wyoming and the group, to promote CS education.

From "How a State Plans to Turn Coal Country into Coding Country"
The New York Times (08/10/19) Amy Goldstein
View Full Article - May Require Paid Subscription

Hackers Told to Break into U.S. Voting Machines Didn't Have Much Trouble

The Washington Post

Hackers at this year's Def Con hacker conference successfully exploited weaknesses in U.S. voting systems, demonstrating that many machines could be hijacked and abused through their Internet connections. Nordic Innovation Labs' Harri Hursti said nearly all of the exploitable equipment showcased at Def Con's Voting Village is still being used in U.S. elections, with vendors continuing to sell them despite the presence of serious flaws. Events at the Voting Village included the demonstration of a $10-million experimental voting system from the U.S. Defense Advanced Research Projects Agency. Local election officials cited a lack of information on the security of state voter registration systems, and Hursti added that vendors have threatened litigation to discourage research into product security, rather than correcting their equipment's weaknesses. Hursti said, "Everyone claiming we can fix this by 2020 is giving a false sense of security."

From "Hackers Told to Break into U.S. Voting Machines Didn't Have Much Trouble"
The Washington Post (08/12/19) Taylor Telford
View Full Article - May Require Paid Subscription

Monitoring the Matterhorn with Millions of Data Points

ETH Zurich

Researchers at ETH Zurich in Switzerland have been using a wireless sensor network on the Matterhorn's Hornli ridge to measure terrain, permafrost, and weather for the last decade, to assemble data on natural dangers like destabilization. The scientists accumulated 115 million data points almost in real time using 17 different sensor types installed at 29 locations, in and around a rockfall zone. The data from sensors was transmitted from the ridge via wireless local-access network to the summit station of the cable car of the Klein Matterhorn, then sent to ETH Zurich's data center for continuous analysis. The researchers removed ambient noise through machine learning and smart algorithms programmed directly into sensors. ETH Zurich's Jan Beutel said, "This dataset constitutes the longest, densest, and most diverse data record in the history of alpine permafrost research worldwide."

From "Monitoring the Matterhorn with Millions of Data Points"
ETH Zurich (08/13/19) Peter Ruegg
View Full Article

New Tools Help Detect Digital Domestic Abuse

Cornell Chronicle (NY)

Cornell University researchers have developed a model that can respond systematically and effectively to the growing range of digital threats against victims of intimate partner violence. The researchers created and piloted a questionnaire, a spyware scanning tool, and a diagram for assessing clients' digital footprints. The model can help counselors without tech experience identify online abuse, while protecting abuse victims and their advisers. The team used this method to find potential spyware, compromised accounts, or exploitable misconfigurations for 23 of 44 clients it advised. Said Cornell researcher Diana Freed, "This is giving people a more accurate way to make decisions and providing them with a comprehensive understanding of how things are happening."

From "New Tools Help Detect Digital Domestic Abuse"
Cornell Chronicle (NY) (08/13/19) Melanie Lefkowitz
View Full Article

Data Assimilation Method Offers Improved Hurricane Forecasting

Penn State News

Researchers at the Pennsylvania State University (Penn State) Center for Advanced Data Assimilation and Predictability Techniques have developed a new approach that successfully predicted the intensity and trajectory of Hurricane Harvey, which struck the Houston metropolitan area in 2017. The approach uses data available at the time from the GOES-16 satellite, paired with Penn State's all-sky radiance model. While the researchers still want to study all other hurricane events with new satellite data, the new method provides hope for the future of hurricane forecasting. Said Penn State's Xingchao Chen, "We will continue to test our satellite data assimilation system with more hurricanes to see if this method works well with other severe weather events."

From "Data Assimilation Method Offers Improved Hurricane Forecasting"
Penn State News (08/15/19) David Kubarek
View Full Article