Association for Computing Machinery
Welcome to the February 5, 2016 edition of ACM TechNews, providing timely information for IT professionals three times a week.

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


Do Computer Scientists Hold the Key to Treating Cancer?
The Huffington Post (02/04/16) David Patterson

Big data analysis could be critical piece of the puzzle for ultimately curing cancer, writes former ACM president David Patterson. If the 1.7 million Americans who will get cancer this year were to have their healthy and tumor cells sequenced, it would generate one exabyte of raw data. The University of California, Berkeley AMPLab, has been collaborating with Microsoft Research and the University of California, Santa Cruz since 2011 in a push for using cloud computing and open source software in the war against the disease. This joint effort has yielded software that filters out the human portion of DNA data, and then sequences the remaining pathogen. Unfortunately, Patterson notes, genetic repositories are still a factor of 10-100 short of having sufficient cancer patients from whom to draw statistically significant results. Big data accessibility is a major factor, and in 2013 the Global Alliance for Genomics and Health was founded "to enable the responsible, voluntary, and secure sharing of genomic and clinical data." To prepare for the day when that data will eventually be shared, Patterson writes, we need to enlist software experts now who can leverage advances in cloud computing and machine learning while shielding patient privacy to start building open source tools that will enable scientists to make major progress on cancer.

Virtual City Walkthroughs Help to Find Pedestrian Death Traps
New Scientist (02/04/16)

Columbia University researchers are engaged in a project that uses home-built software called CANVAS for the purpose of examining Google Street View images to better identify locations in New York City where pedestrians are at high risk. Columbia epidemiologist Andrew Rundle's research team scoured 532 New York City intersections, in search of issues that could be contributing to pedestrian injuries or deaths. They compared their observations against a public database of pedestrians struck by cars, and learned such accidents were more likely at corners with billboards, bus stops, and pedestrian signals. Rundle says the hope is these insights can be used to enhance street safety. For example, the city might try experimenting with changing where buses stop or how they unload passengers, to see if that reduces pedestrian injuries. "Cities produce immense amounts of data, and for a long time, that data that was hard to tap into or locked up in corporate or government hard drives," Rundle notes. "Now, there's enormous potential for all of this information to be harnessed to help understand health conditions."

Querying Historical Maps as a Unified, Structured, and Linked Spatiotemporal Source
CCC Blog (02/02/16) Helen Wright

Yao-Yi Chiang, whose research lies at the intersection of computer science and spatial science, is building techniques to unlock historical information from maps. The University of Southern California researcher says historical maps offer detailed geographic information, but finding relevant maps can be difficult and machines cannot read their content. "I envision a map processing, modeling, linking, and publishing framework that allows querying historical map collections as a unified and structured spatiotemporal source in which individual geographic phenomena [extracted from maps] are modeled with semantic descriptions and linked to other data sources [e.g., DBpedia]," says Chiang. This would enable historical spatiotemporal datasets on a large scale to be studied, both in time and space. Moreover, problems could be solved that can not be easily answered today. Chiang and colleagues have built a computer program that "reads" historical Ordinance Survey maps to quickly generate data "labels" for identifying likely locations of subterranean contamination that would not be known today. His paper, took first place at the Computing Community Consortium-sponsored Blue Sky Ideas Track Competition during the 2015 ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems in Seattle in November.

World's First Single-Atom Optical Switch Fabricated
IEEE Spectrum (02/02/16) Dexter Johnson

Researchers from ETH Zurich Switzerland have developed the equivalent of a single-atom photonic transistor by fabricating the world's first single-atom optical switch. The team leveraged plasmonics to create an optical switch that uses a small voltage to relocate an atom and essentially produce a switch that can be turned on or off. The device features two pads of metal--one silver and one platinum--placed on top of an optical waveguide made of silicon, with only a few nanometers separating them. When the light comes in from an optical fiber, it is guided to the gap between the two metals by an optical waveguide. The light waves cannot pass through, but the plasmon waves on the surface of the metallic pads can. Once the waves of electrons pass through the gap, they can then be turned back again into an optical signal. The team believes the device could be ready for series production in a few years with the introduction of 10-nm lithography.

Novel Semantic Tagging Tool to Benefit Digital Journalism
CORDIS News (02/03/16)

A new project funded by the European Union promises to make digital journalism more effective. Researchers involved in the Media in Context (MICO) project have developed a semantic editor, called WordLift, that structures content with a classification scheme to provide the needed context to news stories. MICO researchers partnered with Greenpeace Italy and Shoof, a startup developing an Android app for user generated content, as part of their effort to create a flexible network of metadata around both text and data for small and medium-sized editorial teams. The researchers report structuring content and creating multiple access points dramatically increases overall content discoverability over social networks and search engines. By using semantic tagging, content publishers begin curating a set of concepts that emerge from the content being produced and analyzed. With WordLift, a plugin for WordPress, these concepts are gathered by applying an internal vocabulary. The tool can provide independent news organizations with the content structure and context needed to engage, capture, and retain target audiences.

Carnegie Mellon Project Aims to 'Reverse-Engineer' the Brain
IDG News Service (02/03/16) Katherine Noyes

On Wednesday, Carnegie Mellon University's (CMU) Computer Science Department and the Center for the Neural Basis of Cognition launched a five-year, $12-million project to "reverse-engineer" the human brain, with the ultimate goal of improving neural networks. "Powerful as they are, [neural networks] still aren't nearly as efficient or powerful as those used by the human brain," notes CMU professor Tai Sing Lee. Researchers will use two-photon calcium imaging microscopy to record the signaling of tens of thousands of individual neurons in mice as they process visual information. "By incorporating molecular sensors to monitor neural activity in combination with sophisticated optical methods, it is now possible to simultaneously track the neural dynamics of most, if not all, of the neurons within a brain region," says CMU professor Sandra Kuhlman. She notes a massive data set will be compiled, providing a fine-grained picture of how neurons in one region of the visual cortex behave. "The hope is that this knowledge will lead to the development of a new generation of machine-learning algorithms that will allow [artificial-intelligence] machines to learn without supervision and from a few examples, which are hallmarks of human intelligence," Lee reports. The U.S. Intelligence Advanced Research Projects Activity is funding the project via its Machine Intelligence from Cortical Networks program.

Clever New GitHub Tool Lets Coders Build Software Like Bridges
Wired (02/03/16) Cade Metz

GitHub engineers have released an open source tool called Scientist designed to ensure new software code is ready before the old code is disconnected. Scientist uses the Branch By Abstraction engineering method, in which it wraps service in an additional abstraction layer of software that manages outside communication, juggling all inputs and outputs. Programmers then can write new code to fit the abstraction layer, guaranteeing it can handle all the same inputs and outputs. Once this is done and properly tested, the coder can flip a switch so the abstraction layer points not to their old code, but to the new. The abstraction layer can run the old and new code in parallel, with the same live data streaming into both systems, while Scientist records any behavioral differences. Using this data, the coder can make any needed refinements to the new code. "It makes sure [code changes] are done safely and that they're not destroying what's there already or introducing new bugs," says GitHub engineer Jesse Toth. Although Scientist is designed to work with GitHub's underlying programming language, Ruby, Toth says the same techniques are applicable to any other language.

The Nanodevice Aiming to Replace the Field Effect Transistor
Technology Review (02/02/16)

University of North Carolina in Charlotte researcher Jason Marmon and colleagues recently unveiled a light effect transistor as a possible replacement for the field effect transistor. The device basically consists of a nanowire that conducts when it is exposed to light and insulates when it is dark. The research team says it is simpler than a field effect transistor and does not rely on dopant atoms, so it can be miniaturized and thus continue Moore's law. They report these wires, fashioned from cadmium and selenium, exhibit useful and unique behaviors, operating as switches that enable a million times more current to flow when they are on compared with off when running at a voltage of about 1.5 V. The wires also are imbued with other capabilities outside those of field effect transistors. The device functions as an optical amplifier and also can execute basic logic operations by using two or more laser beams instead of one. The major benefit is that because the photoconductive effect does not require dopant atoms, it is not vulnerable to the problems of random variation that plague field effect transistors. In addition, nanowires are simpler than field effect transistors, which makes them potentially less expensive and easier to manufacture.

HCRI Advances Technology to Solve Human Problems
Brown Daily Herald (02/03/16) Hattie Xu

Brown University professor Michael Littman and colleagues launched the Humanity-Centered Robotics Initiative (HCRI) in December after receiving new funding and support from the university. The overall aim of HCRI is to investigate and meet societal needs though human-robot interaction. The funding will enable HCRI to grant seed funds to more researchers and to set up a "living room" to study how robots can interact in a home environment. HRCI is led by Littman and Bertram Malle, associate chair of cognitive, linguistic and psychological sciences, who have been working together on human-robot interaction research for about three years. Malle notes other major universities focus on robotics in a technical way, but Brown will address human problems. HCRI will research ways to apply robotics to society in a beneficial way, and focus on ethics, policy, and security. The initiative also will address the challenge of building a moral robot, Malle says. "If you have robots that interact with humans--that increasingly become autonomous and make decisions--those robots need to know about social and moral norms," he notes. "Our goal is to study how the human moral system works and how this knowledge can be transferred to robots."

Hack-Proof RFID Chips
MIT News (02/03/16) Larry Hardesty

Researchers at the Massachusetts Institute of Technology (MIT) and Texas Instruments have developed a virtually hack-proof radio-frequency identification (RFID) chip, which they presented this week at the International Solid-State Circuits Conference in San Francisco. MIT graduate student Chiraag Juvekar says the chip is designed to foil side-channel attacks, which analyze patterns of memory access or fluctuations in power consumption when a device is conducting a cryptographic operation, in order to extract its cryptographic key. The RFID chip's effectiveness in preventing such attacks is courtesy of two design advances: an on-chip power supply whose link to the chip circuitry would be virtually impossible to sever, and an array of "nonvolatile" memory cells that can store whatever data the chip is working on when it starts to lose power. The device utilizes ferroelectric crystals and a bank of 3.3-volt capacitors as an on-chip energy source, while 571 1.5-volt ferroelectric cells are embedded into its circuitry. When the chip's power source, an external scanner, is removed, the chip harnesses the 3.3-volt capacitors and completes as many operations as possible, then stores the data it is working on in the 1.5-volt cells. When power is reintroduced, the chip recharges the capacitors so that if another interruption occurs, it will have sufficient power to store data. It then resumes its previous computation and if that computation was an update of the secret key, it will finish the update before responding to a query from the scanner, thwarting power-glitch attacks.

Robotic Fingers With a Gentle Touch
Ecole Polytechnique Federale de Lausanne (02/01/16) Hillary Sanctuary

Researchers at the Swiss Federal Institute of Technology in Lausanne (EPFL) have developed a new soft robotic gripper that can bend and pick up delicate objects using electroadhesion. When the voltage is turned on, the electrodes bend toward the object to be picked up, imitating muscle function. The tips of the electrodes act like fingertips that gently conform to the shape of the object, gripping it with electrostatic forces. The electrodes can carry 80 times their own weight without requiring any prior knowledge about the object's shape. "Our unique configuration of electrodes and silicone membranes is what allows us to control the bending of the flaps and the electrostatic grip," says EPFL researcher Herbert Shea. The electrode flaps consist of five layers: a pre-stretched elastomer layer sandwiched between two layers of electrodes, plus two outer layers of silicone of different thickness. When the voltage is off, the difference in thickness of the outer layers makes the flaps curl outwards. When the voltage is on, the attraction between the two layers of electrodes straightens out the membranes, mimicking muscle flexion.

Software Adapts Speech to Ambient Noise Level
Fraunhofer Institute (02/16) Meike Hummerich

Fraunhofer Institute for Digital Media Technology (IDMT) researchers have developed ADAPT DRC, new software that significantly improves the intelligibility of speech. ADAPT DRC continually analyzes ambient noise via a microphone, and the speech is adjusted to it in real time. "Our algorithms are able to prioritize certain frequencies and to reinforce, at the right time, precisely those which are particularly disturbed by the ambient noise," says IDMT researcher Jan Rennies-Hochmuth. In addition, the software accounts for the parts of the speech signal that are of different volumes. Speech intelligibility increases particularly when loud parts are systematically subdued and quiet parts are specifically amplified, a technique known as Dynamic Range Compression (DRC). The ADAPT DRC software has already been developed to the point of application maturity and is available to industrial partners. Modern conference equipment and mobile phones already have built-in microphones, meaning they already possess the technology that is necessary to be able to record the ambient noise. Meanwhile, the speaker systems of railway systems or airports would need additional microphones installed before the ADAPT DRC software could be implemented.

A New, Cheap, and Fast IT System Predicts Crimes Better, Organizes Police Shifts
University of Granada (Spain) (01/29/16)

University of Granada researchers have developed an information technology system based on algorithms that can predict how many and what type of crimes are going to be committed during the next police shift. The researchers say their breakthrough marks the first time predictive police methods have been combined with a mathematical patrolling model, and could better organize police patrolling and districting. The research relies on dynamic systems applied to crime models, and it aims to influence the analysis and development of mathematical and statistical models for identifying temporal patterns for criminal acts. The system "is based in a mathematical, multi-criteria algorithm that, considering multiple performance attributes, assigns the patrol a certain surveillance area, thus preventing the commission of crimes in the next shift, based on the crime risk prediction for that area," according to the researchers. They say the tool strengthens crime prevention, not crime suppression. However, the researchers note the tool only establishes the patrolling area and distributes police officers quickly, and it cannot be a substitute for police experience and intuition.

Abstract News © Copyright 2016 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe