Friday, August 17, 2018

New GAZELLE system that provide security to cloud-based machine learning

A novel encryption method devised by MIT researchers secures data used in online neural networks, without dramatically slowing their runtimes, which holds promise for medical-image analysis using cloud-based neural networks and other applications. Image: Chelsea Turner

MIT scientists have recently developed a novel system that offers security to online neural networks. Scientists dubbed this system as GAZELLE, that blends two conventional techniques.
Using these techniques, homomorphic encryption, and garbled circuits, the system helps the networks run orders of magnitude faster than they do with conventional approaches.

Scientists believe that the system would be helpful for cloud-based neural networks for medical image analysis and other applications that use sensitive data. In addition, it could be used to train CNNs to diagnose diseases.
Scientists tested the system on two-party image-classification tasks. A user sends encoded picture information to an online server assessing a CNN running on GAZELLE. After this, the two both parties i.e., sender and receiver share encrypted information forward and backward with a specific end goal to order the user’s image.
All through the procedure, the system guarantees that the server never adapts any transferred information, while the user never learns anything about the system parameters. Contrasted with customary systems, be that as it may, GAZELLE ran 20 to 30 times speedier than best in class models, while diminishing the required system transmission capacity by an order of magnitude.
First author Chiraag Juvekar, a Ph.D. student in the Department of Electrical Engineering and Computer Science (EECS) said, “In this work, we show how to efficiently do this kind of secure two-party communication by combining these two techniques in a clever way. The next step is to take real medical data and show that, even when we scale it for applications real users care about, it still provides acceptable performance.”
The encryption technique used in the system, i.e, homomorphic encryption, usually uses in cloud computing. It receives and executes computation all in encrypted data, called ciphertext, and generates an encrypted result that can then be decrypted by a user. When applied to neural networks, this technique is particularly fast and efficient at computing linear algebra.
On the other hand, Garbled circuits are a form of secure two-party computation. It takes an input from both parties, does some computation, and sends two separate inputs to each party. In that way, the parties send data to one another, but they never see the other party’s data, only the relevant output on their side.
In their system, a user will transfer ciphertext to a cloud-based CNN. The user must have garbled circuits. The CNN does all the calculation in the linear layer, at that point sends the information to the nonlinear layer. By then, the CNN and user share the information. The user does some calculation on garbled circuits and sends the information back to the CNN.
By part and sharing the workload, the system confines the homomorphic encryption to doing complex math one layer at a given moment, so information doesn’t turn out to be excessively noisy. It likewise restrains the correspondence of the garbled circuits to only the nonlinear layers, where it performs ideally.
The final step was ensuring both homomorphic and garbled circuit layers maintained a common randomization scheme, called “secret sharing.” In this scheme, data is divided into separate parts that are given to separate parties. All parties synch their parts to reconstruct the full data.
Juvekar said, “At the end of the computation, we want the first party to get the classification results and the second party to get absolutely nothing. Additionally, “the first party learns nothing about the parameters of the model.”
Co-authors on the paper are Vinod Vaikuntanathan, an associate professor in EECS and a member of the Computer Science and Artificial Intelligence Laboratory, and Anantha Chandrakasan, dean of the School of Engineering and the Vannevar Bush Professor of Electrical Engineering and Computer Science.
The paper is presented at this week’s USENIX Security Conference.

Thursday, August 16, 2018

Scientists captured Sprawling galaxy cluster hiding in plain sight

An X-ray image (in blue) with a zoom in optical image (gold and brown) showing the central galaxy of a hidden cluster, which harbors a supermassive black hole.

Scientists at the MIT have captured a sprawling new galaxy cluster hiding in plain sight. They found that the cluster which is mere 2.4 billion light-years away from Earth, holds a number of galaxies which, all told, is about as massive as 690 trillion suns. Moreover, scientists estimated that the cluster is surrounded by an extremely active supermassive black hole or quasar. Scientists have published their paper in the Astrophysical Journal.
Scientists dubbed this quasar as PKS1353-341. It is bright enough that it has obscured hundreds of galaxies clustered around it.
During the study, scientists also calculated that the quasar at the center of the cluster is 46 billion times brighter than the sun.
Its outrageous glow is likely the consequence of a transitory encouraging free for all: As a tremendous circle of material swirls around the quasar, enormous pieces of matter from the plate are falling in and sustaining it, making the dark gap transmit gigantic measures of energy out as light.
Study author Michael McDonald said, “This might be a short-lived phase that clusters go through, where the central black hole has a quick meal, gets bright, and then fades away again. This could be a blip that we just happened to see. In a million years, this might look like a diffuse fuzzball.”
In 2012, McDonald and others discovered the Phoenix cluster, one of the most massive and luminous galaxy clusters in the universe. The mystery to McDonald was why this cluster, which was so intensely bright and in a region of the sky that is easily observable, hadn’t been found before.
McDonald said, “We started asking ourselves why we had not found it earlier, because it’s very extreme in its properties and very bright. It’s because we had preconceived notions of what a cluster should look like. And this didn’t conform to that, so we missed it.”
“For the most part, astronomers have assumed that galaxy clusters look “fluffy,” giving off a very diffuse signal in the X-ray band, unlike brighter, point-like sources, which have been interpreted as extremely active quasars or black holes.”
“The images are either all points or fluffs, and the fluffs are these giant million-light-year balls of hot gas that we call clusters, and the points are black holes that are accreting gas and glowing as this gas spirals in. This idea that you could have a rapidly accreting black hole at the center of a cluster — we didn’t think that was something that happened in nature.”
In any case, the Phoenix disclosure demonstrated that galaxy clusters could for sure host tremendously dynamic black holes, inciting McDonald to ponder: Could there be other adjacent universe groups that were basically misidentified?
To answer that inquiry, the analysts set up an overview named CHiPS, for Clusters Hiding in Plain Sight, which is intended to reexamine X-beam pictures taken previously.
For each point source that was beforehand recognized, the analysts noticed their directions and afterward considered them all the more straightforwardly utilizing the Magellan Telescope, a ground-breaking optical telescope that sits in the mountains of Chile. In the event that they watched a higher-than-anticipated number of systems encompassing the point source (a sign that the gas may originate from a group of universes), the analysts took a gander at the source once more, utilizing NASA’s space-based Chandra X-Ray Observatory, to recognize a broadened, diffuse source around the principle point source.
McDonald said, “Some 90 percent of these sources turned out to not be clusters. But the funny thing is, the small number of things we are finding is sort of rule-breakers.”
“The brightness of the black hole might be related to how much it’s eating. This is thousands of times brighter than a typical black hole at the center of a cluster, so it’s very extreme in its feeding. We have no idea how long this has been going on or will continue to go on. Finding more of these things will help us understand, is this an important process, or just a weird thing that there’s only one of in the universe.”
McDonald and his colleagues believe the discovery of this hidden cluster shows there may be other similar galaxy clusters hiding behind extremely bright objects that astronomers have miscatalogued as single light sources. The researchers are now looking for more hidden galaxy clusters, which could be important clues to estimating how much matter there is in the universe and how fast the universe is expanding.
The paper’s co-authors include lead author and MIT graduate student Taweewat Somboonpanyakul, Henry Lin of Princeton University, Brian Stalder of the Large Synoptic Survey Telescope, and Antony Stark of the Harvard-Smithsonian Center for Astrophysics.

Saturday, August 04, 2018

Astronomers detected giant rogue planet bumbling around space

Artist's conception of SIMP J01365663+0933473, an object with 12.7 times the mass of Jupiter, but a magnetic field 200 times more powerful than Jupiter's. This object is 20 light-years from Earth. Credit: Caltech/Chuck Carter; NRAO/AUI/NSF
Astronomers using the National Science Foundation’s Karl G. Jansky Very Large Array (VLA) have discovered a new evidence of a giant rogue planet outside our solar system. The planet which is expected to 12 times larger than Jupiter, found traveling without any sort of set orbit or parent star.
From the radio astronomy observatory, scientists were able to pick up its magnetic activity and study it. Moreover, it has a surprisingly strong magnetic powerhouse and a “rogue”.
Melodie Kao, who led this study while a graduate student at Caltech said, “This object is right at the boundary between a planet and a brown dwarf, or ‘failed star,’ and is giving us some surprises that can potentially help us understand magnetic processes on both stars and planets.”
The strange object in the latest study, called SIMP J01365663+0933473, has a magnetic field more than 200 times stronger than Jupiter’s. The object was originally detected in 2016 as one of five brown dwarfs the scientists studied with the VLA to gain new knowledge about magnetic fields and the mechanisms by which some of the coolest such objects can produce strong radio emission. Brown dwarf masses are notoriously difficult to measure, and at the time, the object was thought to be an old and much more massive brown dwarf.
Last year, another team of scientists found that the SIMP J01365663+0933473 was part of a very young group of stars. It means, it was in fact so much less massive that it could be a free-floating planet — only 12.7 times more massive than Jupiter.
Scientists estimated that the temperature of the object is about 825 degrees Celsius or more than 1500 degrees Fahrenheit.
At the same time, the Caltech group that initially recognized its radio emission in 2016 had watched it again in another examination at significantly higher radio frequencies and affirmed that its magnetic field was much more grounded than first estimated.
Kao said, “When it was announced that SIMP J01365663+0933473 had a mass near the deuterium-burning limit, I had just finished analyzing its newest VLA data. The VLA observations provided both the first radio detection and the first measurement of the magnetic field of a possible planetary mass object beyond our Solar System.”
“This particular object is exciting because studying its magnetic dynamo mechanisms can give us new insights on how the same type of mechanisms can operate in extrasolar planets — planets beyond our Solar System. We think these mechanisms can work not only in brown dwarfs but also in both gas giant and terrestrial planets.”
Gregg Hallinan, of Caltech, said, “Such a strong magnetic field “presents huge challenges to our understanding of the dynamo mechanism that produces the magnetic fields in brown dwarfs and exoplanets and helps drive the auroras we see,”
“Detecting SIMP J01365663+0933473 with the VLA through its auroral radio emission also means that we may have a new way of detecting exoplanets, including the elusive rogue ones not orbiting a parent star.”

Friday, July 27, 2018

New chip enlightens optical neural network demo

NIST’s grid-on-a-chip distributes light signals precisely, showcasing a potential new design for neural networks. The three-dimensional structure enables complex routing schemes, which are necessary to mimic the brain. Light could travel farther and faster than electrical signals. Credit: Chiles/NIST


Using AI, many researchers are trying to emulate the brain by creating circuits of artificial neural networks. But the problem lies in conventional electronics, for example, electrical wiring of semiconductor circuits often obstruct complex routing required for using neural networks.



Researchers at the National Institute of Standards and Technology (NIST) have made a silicon chip that distributes optical signals precisely across a miniature brain-like grid, showcasing a potential new design for neural networks.


The NIST team proposes to use light instead of electricity as a signaling medium. Neural networks already have demonstrated remarkable power in solving complex problems, including rapid pattern recognition and data analysis. The use of light would eliminate interference due to electrical charge, and the signals would travel faster and farther.


NIST physicist Jeff Chiles said, “Light’s advantages could improve the performance of neural nets for scientific data analysis such as searches for Earth-like planets and quantum information science, and accelerate the development of highly intuitive control systems for autonomous vehicles.”
The chip conquers a noteworthy challenge to the utilization of light signals by vertically stacking two layers of photonic waveguides—structures that keep light into thin lines for routing optical signals. This three-dimensional (3D) design enables complex routing schemes, which are necessary to mimic neural systems. Furthermore, this design can easily be extended to incorporate additional waveguiding layers when needed for more complex networks.
These photonic waveguides made from silicon nitride of size 800 nanometers (nm) wide and 400 nm thick. They form a 3D grid with 10 inputs or “upstream” neurons each connecting to 10 outputs or “downstream” neurons, for a total of 100 receivers.
For automated signal routing, scientists have demonstrated a software that adjusts the levels of connectivity between the neurons.
Power levels represent the pattern and degree of connectivity in the circuit. The authors demonstrated two schemes for controlling output intensity: uniform (each output receives the same power) and a “bell curve” distribution (in which middle neurons receive the most power, while peripheral neurons receive less).
During an experiment, scientists created the images of output signals. All signals were focused through a microscope lens onto a semiconductor sensor and processed into image frames. This method allows many devices to be analyzed at the same time with high precision. The output was highly uniform, with low error rates, confirming precise power distribution.The study is published in the journal APL Photonics.

First successful test of Einstein’s general relativity near supermassive black hole

This artist’s impression shows the path of the star S2 as it passes very close to the supermassive black hole at the centre of the Milky Way. As it gets close to the black hole the very strong gravitational field causes the colour of the star to shift slightly to the red, an effect of Einstein’s general thery of relativity. In this graphic the colour effect and size of the objects have been exaggerated for clarity.

F
or the first time, scientists have observed the impacts anticipated by Einstein’s general relativity on the movement of a star going through the extreme gravitational field close to the supermassive black hole in the center of the Milky Way. The study allowed astronomers to follow the star, called S2, as it passed near the black hole amid May 2018.

The star is located at a distance of fewer than 20 billion kilometers from the black hole, moving at a speed in excess of 25 million kilometers per hour. Using the observations made by exquisitely sensitive GRAVITYSINFONI and NACO instruments on ESO’s Very Large Telescope (VLT), scientists can precisely make Newtonian predictions in excellent agreement with the predictions of general relativity.
They compared the position and speed estimations from GRAVITY and SINFONI individually, alongside past observations of S2 utilizing different instruments, with the predictions of Newtonian gravity, general relativity and different hypotheses of gravity. They thus now came up with the culmination of a 26-year arrangement of perpetually exact observations of the center of the Milky Way utilizing ESO instruments.

Their measurements revealed an effect called gravitational redshift. Light from the star is extended to longer wavelengths by the exceptionally solid gravitational field of the black hole. Furthermore, the adjustment in the wavelength of light from S2 concurs unequivocally with that anticipated by Einstein’s theory of general relativity. This is the first time when that this deviation from the forecasts of the simpler Newtonian theory of gravity has been seen in the movement of a star around a supermassive black hole.
The GRAVITY instrument in the VLT Interferometer has tracked the motion of the star S2 as it made a very close approach to the black hole at the centre of the Milky Way. This image shows the star and black hole shortly before their closest approach in May 2018.
Reinhard Genzel of the Max Planck Institute for Extraterrestrial Physics (MPE) in Garching, Germany said, “This is the second time that we have observed the close passage of S2 around the black hole in our galactic center. But this time, because of much-improved instrumentation, we were able to observe the star with unprecedented resolution. We have been preparing intensely for this event over several years, as we wanted to make the most of this unique opportunity to observe general relativistic effects.”
In order to create precise measurements of the changing position of S2 in order to define the shape of its orbit, scientists used SINFONI and quantified the velocity of S2 towards and away from Earth. They also used GRAVITY instrument in the VLT Interferometer (VLTI) that unveils the motion of the star from night to night as it passes close to the black hole. his long-sought result represents the climax of a 26-year-long observation campaign using ESO’s telescopes in Chile.
This simulation shows the orbits of stars very close to the supermassive black hole at the heart of the Milky Way. One of these stars, named S2, orbits every 16 years and is passing very close to the black hole in May 2018. This is a perfect laboratory to test gravitational physics and specifically Einstein’s general theory of relativity.

Françoise Delplancke, head of the System Engineering Department at ESO, explains the significance of the observations: “Here in the Solar System, we can only test the laws of physics now and under certain circumstances. So it’s very important in astronomy to also check that those laws are still valid where the gravitational fields are very much stronger.”
Scientists are further planning to continue their exploration in order to unveil another relativistic effect very soon.
Xavier Barcons, ESO’s Director General, concludes: “ESO has worked with Reinhard Genzel and his team and collaborators in the ESO Member States for over a quarter of a century. It was a huge challenge to develop the uniquely powerful instruments needed to make these very delicate measurements and to deploy them at the VLT in Paranal. The discovery announced today is the very exciting result of a remarkable partnership.”
This diagram shows the motion of the star S2 around the supermassive black hole at the centre of the Milky Way. It was compiled from observations with ESO telescopes and instruments over a period of more than 25 years. The star takes 16 years to complete one orbit and was very close to the black hole in May 2018. Note that the sizes of the black hole and the star are not to scale.
These extremely precise measurements were made by an international team led by Reinhard Genzel of the Max Planck Institute for Extraterrestrial Physics (MPE) in Garching, Germany, in conjunction with collaborators around the world, at the Paris Observatory–PSL, the Université Grenoble Alpes, CNRS, the Max Planck Institute for Astronomy, the University of Cologne, the Portuguese CENTRA – Centro de Astrofisica e Gravitação and ESO.