MIT study reveals new technique for machine-learning systems’ decisions

October 31, 2016

In recent years, the best-performing systems in artificial-intelligence research have come courtesy of neural networks, which look for patterns in training data that yield useful predictions or classifications. A neural net might, for instance, be trained to recognize certain objects in digital images or to infer the topics of texts.

But neural nets are black boxes. After training, a network may be very good at classifying data, but even its creators will have no idea why. With visual data, it’s sometimes possible to automate experiments that determine which visual features a neural net is responding to. But text-processing systems tend to be more opaque. Read more

New tool detects malicious websites before they cause harm

October 27, 2016

Malicious websites promoting scams, distributing malware and collecting phished credentials pervade the web. As quickly as we block or blacklist them, criminals set up new domain names to support their activities. Now a research team including Princeton University computer science professor Nick Feamster and recently graduated Ph.D. student Shuang Hao has developed a technique to make it more difficult to register new domains for nefarious purposes.

In a paper presented at the 2016 ACM Conference on Computer and Communications Security on Oct. 27, the researchers describe a system called PREDATOR that distinguishes between legitimate and malicious purchasers of new websites. In doing so, the system yields important insights into how those two groups behave differently online even before the malicious users have done anything obviously bad or harmful. These early signs of likely evil-doers help security professionals take preemptive measures, instead of waiting for a security threat to surface. Read more

Wearable artificial vision device shows promise in helping legally blind people ‘read’

October 18, 2016

A unique wearable artificial vision device may help people who are legally blind “read” and recognize faces. It may also help these individuals accomplish everyday tasks with significantly greater ease than using traditional assistive reading devices, suggests a study presented today at AAO 2016, the 120th annual meeting of the American Academy of Ophthalmology.

Approximately 246 million people worldwide have low vision. This sight loss impairs a person’s ability to do simple daily tasks. Optical and electronic devices such as hand-held magnifiers, tele-microscopic glasses and computer and video magnifiers can help. But, typically these devices are bulky, cumbersome or not readily portable. With recent advancements in wearable electronic devices and optical character recognition technology that converts images to computer-readable text, University of California, Davis researchers hypothesized that these newer technologies could help improve patients’ ability to function in daily life. To test their theory, researchers asked a group of visually impaired patients to use a wearable artificial vision device to see its impact. They found that the device vastly improved patients’ daily productivity. Read more

T-rays will ‘speed up’ computer memory by a factor of 1000

October 13, 2016

Together with their colleagues from Germany and the Netherlands, scientists at the Moscow Institute of Physics and Technology (MIPT) have found a way to significantly improve computer performance. In their paper published in Nature Photonics, they propose the use of the so-called T-waves, or terahertz radiation as a means of resetting computer memory cells. This process is several thousand times faster than magnetic-field-induced switching.

“We have demonstrated an entirely new way of controlling magnetization, which relies on short electromagnetic pulses at terahertz frequencies. This is an important step towards terahertz electronics. As far as we know, our study is the first to make use of this mechanism to trigger the oscillations of magnetic subsystems,” says Anatoly Zvezdin of Prokhorov General Physics Institute and MIPT, a coauthor of the paper and a USSR State Prize-winning scientist heading MIPT’s Laboratory of Physics of Magnetic Heterostructures and Spintronics for Energy-Saving Information Technologies. Read more

Projection-type see-through holographic 3-D display technology deveoped

October 13, 2016

Summary

- A specially designed optical screen was fabricated using a hologram printer that was developed by NICT.

- Holographic 3D image reconstruction was successfully demonstrated on a see-through screen.

- This technology has potential to be used for in-car head-up displays, smart glasses and digital signage.

The National Institute of Information and Communications Technology (NICT, President: Dr. Masao Sakauchi) has newly developed a projection-type see-through holographic 3D display technology, in which an optical screen of a digitally designed holographic optical element (DDHOE) and a digital holographic projection technique were combined. Holographic 3D image reconstruction was successfully demonstrated via the see-through screen to a target observation area.

Basically, dynamic holographic 3D display technology faces the severe limitation of the spatial-temporal resolution of the spatial light modulator (SLM) to realize a practical display size and visual angle. Moreover, the general system design of holographic 3D displays requires a large optical setup behind the display window.

In this work, a holographic 3D image was largely projected to a see-through screen of DDHOE by using the digital holographic projection technique. The screen was fabricated using a hologram printer, which was uniquely developed by NICT. The light of the enlarged holographic 3D image was then concentrated to a target observation area by an appropriately designed reflection function to increase the visual angle. Because this technology can have a high degree of freedom of both the display size and the visual angle independently, and also a high usability of the see-through display system, it will accelerate the adoption of holographic three-dimensional displays in industrial applications, such as digital signage, in-car head-up displays, smart-glasses and head-mounted displays.

A part of this work was published in Nature Communications on October 3, 2016. This work was supported by JSPS KAKENHI (Grant nos. 26790064 and 16H01742), the MIC/SCOPE (Grant no 162103005) and the Centre of Innovation Program from the Japan Science and Technology Agency, JST.

[Background]

Thanks to the advancements of high-speed computing and communication technology, we can now deal with huge amounts of data, and thus a display technology that finally shows output information is becoming increasingly important. In particular, the demand for a dynamic 3D display technology is increasing with the advancement of 3D printing/scanning technologies. At NICT, several dynamic holographic 3D displays have been developed. However, general holographic 3D displays suffer a trade-off between the display size and the visual angle to view the entire display area owing to the severe limitation of the spatial-temporal resolution of current SLMs.

[Achievements]

NICT has newly developed a projection-type see-through holographic 3D display technology that allows us to design freely both the display size and the visual angle independently of see-through holographic 3D image reconstruction.

For the purpose of creating a new optical technology, the Electromagnetic Application (EMA) laboratory in NICT started the holographic printing technology (“HOPTECH”) project in 2014. As a part of this project, a hologram printer for recording a digitally designed wavefront was developed. This printer can not only print 3D data for 3D visualization but also fabricate a holographic optical element (HOE) with an arbitral reflection function (we call this kind of hologram a digitally designed holographic optical element (DDHOE)).

A new approach was achieved using two key technologies: digital holographic projection to enlarge the display size and fabrication of an optical screen of DDHOE that reflects the light of a projected holographic 3D image to a target observation area to increase the visual angle. A digital holographic projector was developed by modifying NICT’s previous holographic 3D display system with a projection lens. The appropriate reflection function, namely, the function to concentrate the light of a projected holographic 3D image to the target observation area, of DDHOE was once calculated as wavefront information (amplitude and phase distribution of the light) considering the projection parameters. The wavefront of that reflection function was then implemented into a holographic recording film by using a hologram printer. In comparison with conventional approaches that use a concave mirror or a convex lens to increase the visual angle of a holographic 3D image, the new approach can achieve a higher degree of freedom of display size and visual angle. These parameters are important for practical display use, by using only a thin film of DDHOE.

Additionally, this film has high transparent characteristics at visible light wavelength; this work will open up the real holographic 3D display of some applications such as in-car head-up displays, smart glasses, head-mounted displays and digital signage.

[Future work]

NICT will optimize the full-color display system by upgrading the hologram printer. Multiple holographic projections from different incident directions to the DDHOE will also be focused on to extend the observation area.

High-tech imaging uncovers precolonial Mexican manuscript hidden from view for 500 years

August 19, 2016

Researchers from the University of Oxford’s Bodleian Libraries and from universities in the Netherlands have used high-tech imaging to uncover the details of a rare Mexican codex dating from before the colonization of the Americas. The newly revealed codex, or book, has been hidden from view for almost 500 years, concealed beneath a layer of plaster and chalk on the back of a later manuscript known as the Codex Selden, which is housed at the Bodleian Libraries. Scientists have used hyperspectral imaging to reveal pictographic scenes from this remarkable document and have published their findings in the Journal of Archaeological Science: Reports. Read more

Wi-fi from lasers

August 15, 2016

A nanocrystalline material that rapidly makes white light out of blue light has been developed by KAUST researchers.

While Wi-Fi and Bluetooth are now well established technologies, there are several advantages gained by shortening the wavelength of the electromagnetic waves used for transmitting information.

So-called visible-light communication (VLC) makes use of parts of the electromagnetic spectrum that are unregulated and is potentially more energy-efficient. VCL also offers a way to combine information transmission with illumination and display technologies–for example, using ceiling lights to provide internet connections to laptops. Read more

A new bio-ink for 3-D printing with stem cells

June 23, 2016

The new stem cell-containing bio ink allows 3D printing of living tissue, known as bio-printing.

The new bio-ink contains two different polymer components: a natural polymer extracted from seaweed, and a sacrificial synthetic polymer used in the medical industry, and both had a role to play.

The synthetic polymer causes the bio-ink to change from liquid to solid when the temperature is raised, and the seaweed polymer provides structural support when the cell nutrients are introduced. Read more

Teaching computers to understand human languages

May 9, 2016

Researchers at the University of Liverpool have developed a set of algorithms that will help teach computers to process and understand human languages.

Whilst mastering natural language is easy for humans, it is something that computers have not yet been able to achieve. Humans understand language through a variety of ways for example this might be through looking up it in a dictionary, or by associating it with words in the same sentence in a meaningful way.

The algorithms will enable a computer to act in much the same way as a human would when encountered with an unknown word. When the computer encounters a word it doesn’t recognise or understand, the algorithms mean it will look up the word in a dictionary (such as the WordNet), and tries to guess what other words should appear with this unknown word in the text. Read more

Commonly used strategy for website protection has flaws

April 12, 2016

Cloud-based security providers commonly use DNS redirection to protect customers’ websites. The success of this strategy depends on shielding the website’s original IP address. Computer scientists from KU Leuven, Belgium, and digital research centre iMinds have now revealed that the IP address can be retrieved in more than 70% of the cases. This means that the DNS redirection security mechanism can easily be bypassed.

Websites and online services increasingly have to deal with acts of cybercrime such as ‘distributed denial-of-service’ (DDoS) attacks: the site or service is deliberately bombarded with huge numbers of malicious communication requests from different computers so that it collapses.

“Website owners can protect themselves against cyberattacks by installing dedicated hardware,” says Thomas Vissers from the KU Leuven Department of Computer Science and iMinds. “Yet, this is typically too expensive and too complex for most of them. That’s why website owners often rely on the services offered by cloud-based security providers. One strategy these providers commonly use to protect websites includes diverting incoming web traffic via their own infrastructure, which is sufficiently robust to detect and absorb cyberattacks. However, the success of this strategy heavily depends on how well the website’s original IP address can be shielded. If that IP address can be retrieved, protection mechanisms can easily be bypassed.” Read more

« Previous PageNext Page »