ACM TechNews

RSS

Timely Topics for IT Professionals

ACM TechNews Home

Recent News

The Human Brain Project Reboots: A Search Engine for the Brain Is in Sight

IEEE Spectrum

The European Commission-funded Human Brain Project (HBP) is on track to create a searchable model of the brain by building an infrastructure combining high-performance computing, data analytics, and simulation and modeling software. The effort follows a two-year revamp of the project that shifted HBP's focus from simulation to detailed brain-mapping. Neuroinformatics, high-performance analytics, and computing projects form the project's core, and teams concentrating in these areas plan to deliver software for researchers to access, share, and analyze all types of brain data. A key objective is enabling scientists to integrate datasets to uncover insights about the brain, with one team building a multilevel brain atlas derived from imaging data. Timo Dickscheid at the Julich Research Center in Germany says curating data to ensure currently incompatible datasets can be collated will be a major challenge. Improving computing by applying insights from the brain is one of HPB's goals, with an example being virtual robot systems controlled by cloud simulation.

From "The Human Brain Project Reboots: A Search Engine for the Brain Is in Sight"
IEEE Spectrum (06/21/17) Megan Scudellari
View Full Article

Security Experts Warn Lawmakers of Election Hacking Risks

ZDNet

More than 100 security researchers and computer science experts on Wednesday published a letter warning U.S. lawmakers of the insufficiency of efforts to ensure the integrity of state and federal elections. Signatories offered recommendations that would "form the basis of robust, enforceable, sensible federal standards that can restore needed confidence in American elections," which include guaranteeing that any electronic election machines, or direct recording electronic systems, generate a voter-verified paper ballot to establish the "official record of voter intent." The experts are urging Congress to allocate funds to upgrade state technologies and replace paperless voting systems to "include a good old-fashioned paper ballot, allowing a physical record of a vote that's out of reach from cyberattacks." The letter also calls for creating safeguards against Internet-related security vulnerabilities and assuring the ability to deter attacks, and requiring robust statistical post-election audits before certification of the final results in federal elections.

From "Security Experts Warn Lawmakers of Election Hacking Risks"
ZDNet (06/21/17) Zack Whittaker
View Full Article

Intelligent Underground Robot for Urban Environments Is Designed

Charles III University of Madrid (Spain)

Researchers at the Charles III University of Madrid (UC3M) in Spain say they have developed a new kind of autonomous underground robot with intelligent navigation for urban environments. The researchers want to develop an intelligent system for the autonomous excavation of small-diameter, high-gradient tunnels in urban environments. "The use of innovative localization, mapping, and navigation techniques, along with sensors and georadars, will allow them to be adapted to different land surfaces and aid in the analysis of the work environment and decision making in attaining the goals," says UC3M professor Carlos Balaguer. He notes the system, developed as part of the European Union-supported roBot for Autonomous unDerGround trenchless opERations, mapping and navigation (BADGER) project, aims to become a model for excavation technologies because of its high economic and social impact. Balaguer says BADGER incorporates several innovations, including new applications of robotics to an underground environment.

From "Intelligent Underground Robot for Urban Environments Is Designed"
Charles III University of Madrid (Spain) (06/21/17)
View Full Article

Volunteers Teach AI to Spot Slavery Sites From Satellite Images

New Scientist

Researchers at the University of Nottingham in the U.K. have launched a crowdsourcing project designed to identify South Asian brick kilns, which are frequently the site of forced labor, in satellite images. Project leader Kevin Bales says the data will be used to train machine-learning algorithms to automatically recognize brick kilns in satellite images, which could then be used by non-governmental organizations to investigate for possible signs of slavery. To date, volunteers taking part in the project have identified more than 4,000 potential slavery sites. The participants are presented with a series of satellite images taken from Google Earth, and they are asked to click on the parts of images that contain brick kilns. When 15 volunteers identify each of the nearly 400 images in the dataset, Bales says he will teach the machine-learning algorithm to recognize the kilns automatically.

From "Volunteers Teach AI to Spot Slavery Sites From Satellite Images"
New Scientist (06/22/17) Matt Reynolds
View Full Article

Illinois Researchers Build Dropbox-Like Storage, Analytical System for Scientific Data

University of Illinois News Bureau

Researchers at the University of Illinois at Urbana-Champaign (UIUC) are studying how to accelerate the materials-to-device process through a new framework called 4CeeD: Real-Time Data Acquisition and Analysis Framework for Material-related Cyber-Physical Environments. The framework links microscopes and other scientific instruments to a cloud infrastructure via a high-speed campus network. "4CeeD enables researchers to search for experiments with specific parameters and receive insights into their own work," says UIUC professor Klara Nahrstedt. After researchers conduct an experiment using a scientific instrument, they can upload the data, including image files, to the cloud. 4CeeD enables the researchers to tag the files with metadata, which helps them search for information about the experiment later. The system also helps faculty members create a template that students can use when annotating files for similar experiments. 4CeeD won a Best Paper award in May at the IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing (CCGrid 2017) in Madrid, Spain.

From "Illinois Researchers Build Dropbox-Like Storage, Analytical System for Scientific Data"
University of Illinois News Bureau (06/19/17)
View Full Article

NIST Awards $38.5 Million to Accelerate Public Safety Communications Technologies

CCC Blog

The U.S. National Institute of Standards and Technology (NIST) has awarded grants totaling $38.5 million to 33 research and development (R&D) projects committed to advancing broadband communications technologies for first responders. The goal of the NIST grants is to modernize public safety communications and operations by supporting the transfer of data, video, and voice communications from mobile radio to a nationwide public safety broadband network, as well as expediting critical indoor location-tracking and public safety analytics technologies. The awardees encompass five technology areas with the potential to significantly augment public safety communications and operations, including mission-critical voice, location-based services, public safety analytics, research and prototyping platforms, and resilient systems. Among the winning projects are the University of Michigan's Body-Worn Camera Analytics in Public Safety initiative, the University of Virginia's Towards Cognitive Assistant Systems for Emergency Response effort, and Carnegie Mellon University's Real-Time Video Analytics for Situation Awareness project.

From "NIST Awards $38.5 Million to Accelerate Public Safety Communications Technologies"
CCC Blog (06/20/17) Helen Wright
View Full Article

Google Open Sources TensorFlow Training Tools

InfoWorld

Google has open sourced its Tensor2Tensor (T2T) project to reduce the volume of work in configuring the TensorFlow deep-learning model for training. T2T consists of a Python-powered workflow organization library for TensorFlow training tasks, enabling developers to specify the key elements used in a TensorFlow model and define the relationships among them. T2T features embedded support for common training datasets, problems and modalities characterizing the jobs the training is for and what types of data to expect for it and produce from it, common models already registered with T2T with room to add more, hyperparameters, and trainers. The most immediately useful aspect of T2T are built-in default settings for each element, so users can quickly get started by reusing or broadening an existing model and deploy one of the defaults and modify it as needed. However, T2T does not provide a larger context outside of TensorFlow for structuring a deep-learning project.

From "Google Open Sources TensorFlow Training Tools"
InfoWorld (06/22/17) Serdar Yegulalp
View Full Article

How Scratch and Minecraft Developers Hope to Keep Kids Coding for Life

EdSurge (CA)

Researchers at the Massachusetts Institute of Technology's Media Lab aim to nurture a lifelong enthusiasm for programming among young students, using the Scratch learn-to-code platform. "Too often people are positioning Scratch as a stepping stone to other things, whereas we see...that Scratchers keep going deeper and broader and contribute back either online or offline to other people," says Scratch developer Natalie Rusk. Essential to this phenomenon are what Rusk calls "interest-based communities" that keep children engaged and learning to code outside of school. Rusk wants to foster these online communities by permitting them to pursue projects aligned with their hobbies in music, games, and the arts. Meanwhile, Microsoft's Minecraft developers are working to cultivate an open coding community for teachers and students, applying user feedback to inform their Education edition of the popular game. "Adding coding to the platform has brought people who were coding-focused, but not necessarily Minecraft-focused on to the platform," notes Minecraft Education's Neal Manegold.

From "How Scratch and Minecraft Developers Hope to Keep Kids Coding for Life"
EdSurge (CA) (06/21/17) Jenny Abamu
View Full Article

What an AI's Non-Human Language Actually Looks Like

The Atlantic

Researchers at the Facebook Artificial Intelligence (AI) Research lab have found that bots they were training to negotiate with each other began conversing in a non-human language without human supervision. Although visually the bots' language appears completely nonsensical, a lab spokesperson says Facebook's data shows that sometimes such dialogues ultimately lead to successful negotiations. Other AI researchers have reported witnessing machines developing their own languages, including languages with a coherent organization, defined vocabulary, and syntax, although they do not always communicate meaning that is comprehensible to humans. For example, scientists from OpenAI describe bots learning to talk in an abstract language and resorting to non-verbal communication when language communication is unavailable. University of Pennsylvania professor Mark Liberman doubts the Facebook bots' language will have longevity--partly because it is completely textual as opposed to human language's basis in speech and gestures, and partly because they will likely be outdated by future machine-learning algorithms.

From "What an AI's Non-Human Language Actually Looks Like"
The Atlantic (06/20/17) Adrienne LaFrance
View Full Article

Teaching Computer Science Is Great, but It's Not Enough

Education Week

Although progress has been made on providing more computer science (CS) education to K-12 students, the U.S. is still years away from providing adequate (CS) education in all schools so the field can realize its full potential, write University of Massachusetts, Amherst professor Florence R. Sullivan and Education, Training, and Research scientist Jill Denner. They advocate for educators to teach functional computer science literacy, in addition to opening up access to more K-12 students. Sullivan and Denner also cite studies indicating white males dominate the most powerful U.S. tech companies, creating inequity and a lack of diversity. "Because of this, we must teach children not just to think about how to design and program a particular technology, but to consider its potential role and impact on society," according to Sullivan and Denner. For example, they say educators should teach students about the ethical ramifications of creating technologies from few viewpoints versus multiple viewpoints.

From "Teaching Computer Science Is Great, but It's Not Enough"
Education Week (06/20/17) Florence R. Sullivan; Jill Denner
View Full Article

Is It Unethical to Design Robots to Resemble Humans?

Quartz

The growing tendency for people to anthropomorphize machines is being hastened by artificial intelligence (AI) and other technologies' ability to create the illusion of self-awareness. However, there is a perceived danger that humans' objectification of human-looking objects may easily transfer to human relationships, and society's push toward humanization of AI could have the unplanned effect of dehumanizing actual human beings. This outcome raises several issues worth considering, including how society should treat realistic AI. A concern exists that extreme dominance over realistic AI could encourage the abuse of animals and people, which dovetails with the question of how much realism should companies imbue within AI technologies. As virtual reality and AI technologies become increasingly lifelike, issues relating to the visceral impact of realistic virtual experiences will become more prominent. Society should address both the treatment and the development of realistic AI technologies simultaneously, and establish norms of behavior and better guidelines for their development.

From "Is It Unethical to Design Robots to Resemble Humans?"
Quartz (06/22/17) David Ryan Polgar
View Full Article

Shrinking Data for Surgical Training

MIT News

Researchers at the Massachusetts Institute of Technology (MIT) say they have developed a system that can efficiently search through hundreds of hours of video footage for events and visual features that correspond to a few training examples. The researchers say they trained the system to recognize different stages of an operation, such as biopsy, tissue removal, stapling, and wound cleansing. The team says the system also could be applied to any analytics question that is relevant to doctors. For example, it could be trained to predict when particular medical instruments should be prepared for the surgeon's use, or it could sound an alarm if a surgeon encounters rare, aberrant anatomy. "The other thing that is extraordinarily exciting to the surgeons is that in the future, we should be able to monitor the progression of the operation in real time," says MIT professor Daniela Rus.

From "Shrinking Data for Surgical Training"
MIT News (06/16/17) Larry Hardesty
View Full Article

A Unique Data Center for Cosmological Simulations

Ludwig-Maximilian University of Munich

Researchers from the Excellence Cluster Universe at the Ludwig-Maximilian University of Munich (LMU) in Germany have established "Cosmowebportal," a datacenter for cosmological simulations, at the Leibniz Supercomputing Center (LRZ) of the Bavarian Academy of Sciences. The new datacenter provides access to the results of Magneticum Pathfinder, the world's most extensive set of cosmological hydrodynamic simulations. The complete simulations are saved at the LRZ, which is connected to the SuperMUC supercomputer. Other interested researchers can use a Web interface to select objects from the raw simulation data, process it, and generate virtual observations mimicking existing or future space telescopes. Astronomical projects to be launched in the next few years will observe large areas of the universe and provide further insight into the evolution of the first structures of the cosmos. "A datacenter that pools and makes these simulations available therefore is an important facility for scientists working in the field," says LMU researcher Klaus Dolag.

From "A Unique Data Center for Cosmological Simulations"
Ludwig-Maximilian University of Munich (06/15/17) Petra Riedel
View Full Article