Efinix’s Programmable Chips Could Push AI Out to the Edges

Startup gets $9.5 million to develop smaller, more flexible FPGAs

Photo: Yuichiro Chino/Getty Images
By its cofounder’s reckoning, Efinix is in the right place at the right time. Engineers are struggling to squeeze AI, and especially its deep learning variant, into chips where cost and power are real constraints.

The startup, based in Santa Clara, Calif., plans to deliver a new kind of field-programmable gate array (FPGA) technology that is about one-quarter the size of comparable chips, consumes half the power, and is considerably less complex to construct. That combination—what Efinix calls Quantum programmable technology—could help push deep learning and AI in general away from central computers and servers and out toward where the data they work on is being generated, according to cofounder, president, and CEO Sammy Cheung.

FPGA’s have been using the same basic architecture for decades. From a high-level, it looks like a checkerboard, with alternating sections that are dedicated to either routing or logic. Cheung and cofounder, FPGA-expert Tony Ngai, came up with a new scheme. Instead of each square of the board having a dedicated function, each square—called an eXchangeable Logic and Routing cell (XLR)—can be programmed to perform either purpose.

Traditional FPGA routing blocks have to be designed to handle the worst case scenario: the most complex set of interconnections possible. Because of that, modern FPGAs need a full 10 to 14 layers of metal to form all the interconnects. Those layers of metal and their accompanying insulation act as parasitic capacitors, draining away power.

But because the role of each block is flexible in the Quantum programmable tech, the new design doesn’t need to worry about that worst case. If a logic block needs particularly complex routing, all you need to do is assign an extra one of its neighboring blocks to do the routing. That means the Quantum system is inherently smaller and requires only seven layers of metal interconnect. The reduced metal greatly decreases the power draw from parasitic capacitance, while also making it simpler to integrate the architecture into another chip, such as a System-on-Chip or an applications specific integrated circuit (ASIC).

Efinix plans to begin producing products with development partners starting in 2018 thanks to a US $9.5 million round of funding that completed last week. Interestingly, the biggest contributor was leading FPGA-maker Xilinx. “Efinix’s solution can address a wide variety of applications that are typically not served by today’s FPGAs,” Salil Raje, senior vice president of the software and IP products group at Xilinx, said in a press release. “We are excited to be an investor and look forward to working with them.”

“We’re not going to compete with Xilinx,” says Cheung. “Instead, we can expand market for FPGA.” That market is worth about $5 billion today, he says, but it’s not growing very fast. But “we can see a market that’s $10 billion.”

Powered by WPeMatico

Celebrating the life of doctoral student and alumnus Michael B. Cohen

Michael B. Cohen ’14, SM ’16 had a deep love for mathematics and the theoretical foundations of computing — a love that was infectious, brilliant, and always shared with others. Cohen, a doctoral student in the Department of Electrical Engineering and Computer Science (EECS), died suddenly from natural causes in September. He was 25 years of age.

At the time of his passing, Cohen was visiting the University of California at Berkeley, where he had gone to meet with colleagues at the Simons Institute for the Theory of Computing. A member of the Theory of Computing group at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, he had roots in the Washington area.

Daniela Rus, the Andrew (1956) and Erna Viterbi Professor of Electrical Engineering and Computer Science and Director CSAIL, said, “We are all still stunned by the news of the passing of Michael Cohen. Michael was a beloved student at CSAIL, a brilliant colleague in the theory group, and a joyful presence everywhere he went. This is a huge and collective loss for the entire CSAIL community.”

Cohen first came to MIT as an undergraduate student, and he lived in East Campus. He earned his bachelor’s in mathematics in 2014, having skipped his second year at MIT to work at Facebook. He then stayed on at MIT to pursue a graduate degree in computer science.

Scott Aaronson, a professor of computer science at the University of Texas at Austin who taught at MIT from 2007 to 2016, recalled Cohen as a particularly motivated first year student, ready to solve open problems, or basically tackle anything.

Writing on his blog, Aaronson noted that when he met Cohen he realized at once that he “was a freshman who I could — must — talk to like an advanced grad student or professor.” In his class on quantum complexity theory, Cohen had the habit of sitting in the front row and carrying on dialogues with Aaronson, often catching any errors or “unjustified claims.” At the same time, Aaronson was impressed by Cohen’s intellectual humility, as his focus was on understanding and clarifying tough concepts for everyone in the class, not showing off. Such openness led to Cohen “having a huge circle of friends.”

His fellow MIT students described him as “energetic, fun, and supportive,” and admired his irrepressible spirit of “exuberance and generosity.” Cameron Musco, a doctoral student in the same lab as Cohen and a frequent co-author, wrote, “It was impossible to ignore his energy, wonder, and excitement for research, current events, and everything in between.” Cohen was a notable presence on the 5th and 6th floors of the Stata Center at MIT, Musco also recalled, “always … surrounded by a group of friends happy to banter or simply to listen. He was a natural teacher — truly kind, humble, welcoming, positive, and always willing to slow his thoughts for a moment to share his brilliance.”

Cohen was an intellectual tour-de-force beyond the campus as well. He spent a summer at Microsoft Research, where he quickly cemented a reputation for “his larger-than-life personality” and academic brilliance. A statement by the team he worked with at Microsoft read: “Michael was a brilliant mathematician and a rising star in his field. … [H]e made sweeping progress in online learning and online algorithms, two fields he had just recently become acquainted with. In addition to solving five open problems in these areas, he continued his substantial progress on the k-server problem, one of the most celebrated and notoriously difficult challenges in the space of adaptive algorithms.”

Sebastien Bubeck, a researcher in the Theory Group at Microsoft Research who worked alongside Cohen, shared what he called a “typical Michael story,” about when they first met in October 2016 at MIT: “We were about to start lunch with a small group of graduate students and Michael entered the room, he (gently) interrupted the conversation and his first sentence to me was a question about mirror descent that I was not able to answer. (We now know the answer, and as it turns out his question was pretty deep and the answer highly non-trivial.)”

Bubeck, like many others, was also struck by Cohen’s remarkable way of doing mathematics, primarily never writing anything on paper. James R. Lee, a professor of computer science at University of Washington, said, “His mind was always going at 100 mph, so it was remarkable that he didn’t miss a beat in calibrating (i.e., slowing down) for an audience (or for those who did not know him).”

Luca Trevisan, a professor of electrical engineering and computer science at Berkeley, noted that “in a few short years, Michael left his mark on a number of problems.” At the time of his death, Cohen was credited as a co-author on papers with more than 30 distinct collaborators. Tom Cohen, Michael’s father, remarked that his son, “more than anything, wished to become part of that community and to engage in a meaningful way on relevant research in the field.” Cleary, he achieved that and much more, becoming a leading light in the theoretical computer science community.

As Lee wrote, “one got the sense that this was all a warmup for Michael. It’s really disheartening that we won’t get to see what comes next.” He and others have committed to ensuring that Cohen’s work is more than just remembered, but spread far and wide and used to tackle the kinds of open problems that he adored.

Those wishing to making contributions in Cohen’s name should consider givedirectly.org, a charity he admired that provides money to poor people in Africa.

<

p class=”wpematico_credit”>Powered by WPeMatico

    Life 3.0 Artificial emotional intelligence

  • Autopilot Internet Income Ewen Chia's Autopilot Internet Income (Brand New!)
  • great_links_to_your_website Discover How To BOOST Your Website's Link Popularity And PageRank With These Seven Little Known, Battle-Tested, And Easy To Use Linking Tactics -- And What's More... I'll Show You How To Get Up And Running With Nearly All Of Them For FREE

Global Kaposi Sarcoma Market Is Expected To Reach US$ 135.6 Mn By 2025

Global Kaposi Sarcoma Market Is Expected To Reach US$ 135.6 Mn By 2025

The latest market report published by Credence Research, Inc. “Kaposi Sarcoma Market – Growth, Future Prospects, Competitive Analysis, 2017 – 2025,” the global kaposi sarcoma market was valued at US$ 108.8 Mn in 2016, and is expected to reach US$ 135.6 Mn by 2025 expanding at a CAGR of 2.44% from 2017 to 2025.

Browse the full report Kaposi sarcoma Market – Growth, Future Prospects, Competitive Analysis, 2017 – 2025 at http://www.credenceresearch.com/report/kaposi-sarcoma-market

Market Insights

Kaposi sarcoma has an incidence rate of 0.02-0.06 %, it is widely prevalent in middle aged Afro-American men, East-European and Mediterranean region men of Jewish origin (Ashkenazi). Kaposi sarcoma has overwhelming male predominance. Serological assays for the detection of Kaposi sarcoma are limited and has often resulted in the production of conflicted data with different methodologies. Polymerase chain reaction, insitu hybridization, and immunohistochemistry reveal the KSHV proteins expressed in the human tissue.

In 2016, HAART (highly active antiretroviral therapy) segment dominated the market due to rising prevalence of AIDS, as patients suffering with HIV infection are 20,000 times prone to develop Kaposi sarcoma and the proactive government initiative such as United States Presidents Emergency Plan for AIDS Relief (PEPFAR) to fight AIDS-KS.

Download Sample Report @ http://www.credenceresearch.com/sample-request/58687

Chemotherapy will be the fastest growing segment throughout the forecast period 2017-2025 majorly due to factors such as technological advancement in the drug formulation for e.g. paclitaxel nanosomal formulation utilized as a topical treatment for AIDS-KS surpassing the drug biotransformation which is toxic to liver cells. Promising pipeline with excellent bioequivalence studies such as AT-0918 (Cytori Therapeutics, Inc.). Aldoxorubicin (CytRx Corporation) utilizes acid sensitive linker that selectively binds to albumin to allow the cytotoxic payload to accumulate near the tumor and spare the surrounding healthy tissues.

In 2016, North America held the largest revenue share mainly due to factors such as rising prevalence of Kaposi sarcoma, presence of pharmaceutical giants primarily focusing on Kaposi sarcoma treatment and presence of developed research and healthcare institutions. Asia Pacific will be the fastest growing market throughout the forecast period 2017-2025, majorly due to increasing disposable incomes in these regions and supportive regulatory environment for Kaposi sarcoma drugs.

Browse the full report at http://www.credenceresearch.com/report/kaposi-sarcoma-market

Key Market Movements:

Rising prevalence of AIDS-KS

Technological advancement has led to improvement in the drug formulation such as targeted immunotherapy and nanosomes to avoid cytotoxicity to healthy cells

Supportive regulatory environment for Kaposi sarcoma drugs

Affordable reimbursement scenario for the treatment of Kaposi sarcoma in developed countries

Powered by WPeMatico

    Life 3.0 Artificial emotional intelligence

  • Senuke Inferno Senuke Inferno is the complete guide to utilizing the SenukeXcr software to skyrocket your site to the top of Google AND make it Panda/Penguin proof FOREVER!
  • Autopilot Internet Income Ewen Chia's Autopilot Internet Income (Brand New!)

Nobel Prize Awarded for Detection of Gravity Waves

Three share award for building LIGO and hearing black holes collide

Photo: Molly Riley/AFP/Getty Images
Rainer Weiss, Barry Barish, and Kip Thorne, have been awarded the Nobel Physics Prize 2017 for their gravitational wave work.
The three men who won this year’s Nobel Prize in physics opened up a new field of science and helped prove Einstein correct by detecting gravitational waves from a pair of colliding black holes.

Rainer Weiss, a physicist from the Massachusetts Institute of Technology, will receive half the prize, while California Institute of Technology physicists Barry C. Barish and Kip S. Thorne will spilt the other half. The three were awarded for conceiving and creating the Laser Interferometer Gravitational Wave Observatory, or LIGO.

LIGO consists of two facilities, one in Hanford, Wash., and one in Livingston, La. Each is made up of two 4-kilometer-long tunnels at right angles to each other. Scientists fire laser beams down each tunnel and measure their reflections. When a gravity wave passes through Earth, it compresses space by a small amount, like a ripple in a pond, and produces a tiny perturbation in the light. In September 2015, LIGO detected the ripple produced when two massive black holes spiraled into each other, 1.3 billion light years away.

Just last week, a sister detector in Italy called Virgo announced the discovery of another collision, the fourth reported so far. That was the first to be measured by three detectors—which allows scientists to locate the source in the sky and point other telescopes toward the event.

“It’s a new branch of science, gravitational wave astronomy,” says Sheldon Glashow, a professor of physics at Boston University, who himself won the Nobel in 1979 for contributions to the theory that two fundamental forces of physics, the weak nuclear force and the electromagnetic force, interacted.

Glashow says the Nobel committee did a good job of divvying up the prize, which totals 9 million Swedish krona, or roughly $1.1 million. “They recognized two of the pioneers of the search, together with the person who actually made it happen.”

Weiss and another scientist, Ron Drever of the University of Glasgow, had separately come up with the idea to use lasers to detect gravitational waves in the mid-1970s. Weiss and Thorne figured out how they might make a detector, and Drever joined them in the project. The group had trouble gaining funding from the National Science Foundation, and eventually Drever, who died this past March, was forced out of the project. Glashow says it wasn’t until Barish came in, in 1994, that the project finally went ahead. “NSF was about to cancel it, but then they said ‘Get Barry, and he’ll solve the problem,’” Glashow says.

Weiss, he says, was the one who figured out what kind of sensitivity such a detector would need to be able to detect gravitational waves. “He was the guy who knew what they needed. Barish was the guy who made it happen,” Glashow said. Thorne, meanwhile, was the evangelist, convincing scientists and the public that this was a worthwhile endeavor.

The 2015 detection recorded the merger of two black holes, one with 29 times the mass of the sun and the other with 36 times the sun’s mass. Glashow says that was surprising to physicists, who believed most ordinary black holes would be only two or three solar masses, except for the giant ones at the centers of galaxies, which can be thousands or millions of times as massive. Now, he says, scientists have to figure out what would produce these intermediate black holes.

LIGO is currently shut down, and the detectors are being upgraded to make them more sensitive. When they’re put back online, they should be able to detect events twice as far away, which means covering eight times the volume of space.

One project will involve trying to measure the polarization of cosmic background radiation, the signature left over from the Big Bang. That, says Glashow, could tell scientists something about the nature of primordial black holes formed near the beginning of the universe, about which they know very little. Measuring the polarization of the gravity waves produced by the collapse of black holes “tells you the tiny little details of Einstein’s theory,” he says.

And some people—not Glashow, he points out—think that gravitational studies will give hints about the existence of axions, theoretical particles that, if they exist, may help explain dark matter, one of the biggest mysteries in cosmology today.

Whatever LIGO and similar detectors find, they’re opening up a new field of science, Glashow says. “Every time we open a window—radio astronomy, x-ray astronomy—we find things we didn’t expect,” he says. With gravitational wave astronomy, “we’ve found the things we expected, but we’re beginning to find the things we didn’t expect.”

Powered by WPeMatico

    Life 3.0 Artificial emotional intelligence

  • Home Biz Launch Formula The Home Biz Launch Formula product consists of SIX training modules to include case studies, coaching sessions, and real world examples. PLUS, you get all the materials and study guides you will need to absorb the coaching and implement the strategies d
  • Commissionology Only Way Left for Little Guy to Get Rich Online

Point of Sale Terminals Market (POS Market) forecast by 2022 published by leading research firm

Emerging technologies increases the demand for the POS (Point of sale) Terminal Market; convenience and flexibility of POS terminals minimized customers waiting time, improved billing process, maintained customer records safely as per security concerns, data back up and pertaining personal information of customer confidential, may inhibit the global POS terminal market.

Digital payment anticipated the POS terminal market growth valued at approximately US$45.97 Bn in 2016 and expected to reach approximately US$98.27 Bn by 2022. Increasing demand for wireless technologies expected to witness growth at 20% of CAGR in between 2017 and 2022.

Click Here To Brows Report on Point of sale (POS) Terminals Market : http://www.credenceresearch.com/report/point-of-sale-pos-terminals-market

POS terminals transmit encrypted tokens during the payment process; its advanced feature includes inventory management, CRM financials and the proliferation of NFC and EMV-enabled devices may drive the industrial market in the next years.

The geographical distribution of the global POS terminals market considered the regional markets of North America, Europe, Asia Pacific, Latin America and rest of the world.

In regional segmentation North America accounted the largest market share of POS terminal followed with Europe and Asia-Pacific respectively. The North America dominated the POS terminal market with the rising adoption of wireless technology across the hospitality applications, restaurants, automotive shops and grocery stores. However the region of Asia-Pacific surprisingly anticipated the POS terminal market with the growing CAGR of 13% in upcoming future. Rapid advancements in the card acceptance and use of debit cards are expected to serve high growth over the forecast period in Asia-Pacific region; especially in India and China; where supermarkets are owing to grow in the upcoming years. It will drive the global revenue of the POS terminals market.

Increasing governmental support in order to improve the digital transaction with advanced technologies boosted the global market of the POS terminals especially in developed economies like Asia-Pacific region. The growing usage of POS terminals with NFC devices in the industrial and retail sector will expected to drive CAGR of 10% over the forecast period. A sustainable governmental effort to promote the computerized payment instead of traditional transaction with cash register will encourage the POS terminal market in this region.

The competitive market for the POS terminals includes New POS technology, NCR Corporation, NFC Corporation, PAX technology, Veri fone Systems, Ingenico SA, Panasonic Corporation, Toshiba Corporation, Cisco Systems and others.

Click Here To Download Full Report Sample on Point of sale (POS) Terminals Market : http://www.credenceresearch.com/sample-request/57716

Application segment of the POS terminals include healthcare, retail, hospitality, entertainment, warehouses, automotive shops, grocery stores and e-commerce sites. The growing usage of POS terminals in these segments witnessed a significant turnaround over the forecast period.

On the basis of component segment POS terminal market segmented into hardware and software, it simplifies the accounting process and promotes the digital payment platform. Card acceptance increases the merchants to enroll the digital transaction will exhibit remarkable opportunities and growth in the POS terminal market.

An advanced and appropriate POS terminal can simplifies the transaction process and save time and money with efficient customer services, it is highly adopted by retailers and e-Commerce companies as per security concerns and confidential financial information of the customers.

Powered by WPeMatico

    Life 3.0 Artificial emotional intelligence

  • 21 Laws of Persuasion (Taboo) The 21 Laws of Persuasion is a 100+ page .pdf that explains the laws of Influence and Sales Psychology with plenty of detailed instructions, real-world examples, and case studies.If your subscribers are interested in the Power of Persuasion and how to a
  • High Ticket Cash Machines OTO1 Your Complete Blueprint to making high ticket affilate sales

More Teachers, Fewer 3D Printers: How to Improve K–12 Computer Science Education

Tech companies and the U.S. government recently pledged $500 million for STEM programs

Photo: iStockphoto
Last week, the Trump White House announced $200 million in federal funding to improve K–12 computer science education every year. The next day, tech leaders including Amazon, Facebook, Google, Microsoft, and Salesforce added another $300 million to the bag, spread over five years. The goal of that initiative is to support K–12 STEM education, focusing on computer science.

The need for basic computer science education has never been greater. Software and computers drive the economy, aiding mines and farms, as well as retail stores, banks, and healthcare. There are 500,000 computer science job openings in the U.S., spanning every industry and state. That’s more than 10 times the number of students who graduated with computer science degrees last year, according to the nonprofit Code.org, which has been working tirelessly to establish and expand CS access in schools.

The tech industry grumbles about the shortage of qualified workers. Yet fewer than half of American K–12 schools offer computer science classes, according to Code.org. That number goes up to around three-fourths of K–12 schools when you include CS exposure through after-school activities and clubs, according to a 2016 Google-Gallup report.

Newer data on high school Advanced Placement exams collected by College Board does show positive signs. The number of “AP Computer Science A” exam-takers more than doubled from 2012 to 2016, reaching more than 54,000 students. And another 40,000 students took the new “AP Computer Science Principles” exam introduced in 2017. But that’s still a small fraction of the 15.1 million students attending high school this year.

A vast majority of students, parents, and educators are interested in and highly value computer science, the Google-Gallup report found. But state policies and curricula are lagging. Ten states currently have K–12 computer science standards, and 10 others are working on them. Only 33 states count computer science courses toward high school graduation credit. Arkansas is the only state that has made K–12 CS education mandatory. Virginia and Rhode Island are on that path.

Getting to this point has taken three decades of tireless state-level advocacy from Code.org and the Association for Computing Machinery. And there’s still a long way to go. Many states treat CS as an elective rather than a core academic subject like math and chemistry. Plus, most state standards focus on basic computer skills, rather than understanding core computing concepts.

Ideal CS courses should teach computational thinking: logical thinking, abstraction, algorithmic expression, problem decomposition, stepwise fault isolation, and debugging. “Every 21st century citizen needs fluency in computational thinking,” says Ed Lazowska, a computer science professor at the University of Washington. “It is never too early to start learning this. Elementary school kids can and do learn it.”

In fact, it’s crucial to start early. Just like reading and math, starting in kindergarten and learning incrementally builds a foundation for logic, critical thinking, and creativity. These skills are difficult to catch-up on later. Plus, says Code.org’s Chief Academic Officer Pat Yongpradit, reaching students earlier helps get girls and minorities on board. “They build confidence and efficacy around their ability to do computer science,” he says. “You can avoid stereotypes or at least make them aware of them.”

Other countries are stepping up CS education efforts. Israel leads the world, according to a report by the Information Technology & Innovation Foundation (ITIF). Its curriculum emphasizes CS as a science instead of teaching only coding, and the nation has 16.2 times as many students on a per-capita basis as the U.S. taking rigorous high school computer science. The U.K. mandates CS for students aged 5 to 14, but is struggling to train all the teachers it needs. According to the ITIF, Australia, Finland, Denmark, and Singapore are reforming CS education by introducing it to primary school students, adding deep concepts to curricula, and training specialized teachers.

Addressing the teacher shortage should be the number one use for the new funds allocated by the Trump administration, says Mark Stehlik, a computer science professor at Carnegie Mellon University. A lack of qualified teachers is the biggest barrier to CS education in the U.S., he says, and he thinks the problem is going to get worse. An earlier generation of CS educators has started to retire, and he says younger CS graduates “aren’t going into education because they can make twice or more working in the software industry.”

One solution could be to expand the reach of each CS educator through online classes. But “online curricula aren’t going to save the day, especially for elementary and high school,” Stehlik says. “A motivated teacher who can inspire students and provide tailored feedback to them is the coin of the realm here.”

Where the money should not be spent? On hardware and equipment. Laptops, robots, and 3D printers are important, says Code.org’s Yongpradit, “but they don’t make a CS class. A trained teacher makes a CS class. So money should be focused on training teachers and offering robust curriculum.”

Corporations also need to take more seriously their responsibility to donate money and technology, and, perhaps more crucially, provide volunteers who can share their skills and knowledge with students. Some top tech firms like Microsoft already have programs that encourage employees to volunteer-teach high school courses, which typically means spending a couple hours a week delivering the class. But it would be more effective if those employees spent several days at the school, teaching students and also mentoring teachers.

This approach can be a win-win for everyone, Stehlik says: building community and social good in the short term, and in the long term, ensuring a workforce for the companies themselves.

Powered by WPeMatico

    Life 3.0 Artificial emotional intelligence

  • Buzzinar Lead Magnet Kit Buzzinar is a viral traffic getting system and software suite. Attract customers build your list and create sales funnels that go viral!
  • Monetizing Dragon Guru Start making money FROM HOME today!

How Experts Comb Satellite Images for Clues on North Korea’s Nuclear Tests

Satellite imagery analysts tracking North Korea’s nuclear program need more than just technical skills

Image: Planet
Satellite images of North Korean missile testing sites or nuclear reactor facilities can sometimes reveal smears of yellow or befuddling brown objects at certain times of the year. A casual armchair observer might suspect those patterns as indicating something nefarious. But they actually represent a mundane harvest practice—North Korean workers often dry out their corn harvest on pavement before putting the harvested crop into brown sacks.

Growing swarms of commercial satellites have provided new tools for both government spies and independent analysts to peek inside North Korea as the isolated “Hermit Kingdom” races to strengthen its arsenal of nuclear weapons. But the process of understanding satellite images of suspected North Korean missile or nuclear test sites is far from easy and requires more than just the latest satellite imaging technology. The best analysts need more than the skills to analyze near-infrared imagery or use mapping software—they must also tap into a wide swath of cultural and technical knowledge when trying to figure out what a particular satellite image can reveal about the secretive North Korean nuclear program.

“Satellite imagery interpretation is very interdisciplinary,” says Melissa Hanham, a senior research associate in the East Asia Nonproliferation Program at the Middlebury Institute of International Studies
(MIIS) at Monterey, Calif. “You have to be a little bit of an expert in everything, and you have to know your limits.”

Analysts have access to more satellite data than ever these days. Companies such as DigitalGlobe and Airbus have satellites capable of providing high-resolution images of ground objects as small as a third of a meter per pixel for commercial customers. Newer companies such as Planet have deployed constellations of dozens of smaller satellites with lower resolution capabilities around 3 meters per pixel. But the swarm of small satellites can capture more frequent images of locations in North Korea and elsewhere, which provides analysts a new tool for tracking a given site’s normal “pattern of life.”

Both increased access to daily satellite imagery and broad expertise on North Korea proved helpful in a 2016
Washington Post 
analysis by Hanham and her colleagues. Part of their analysis involved identifying corn piled along the access road leading to Panghyon Airbase in Kusong, North Korea, which represents an alternative missile test site for the regime. They were able to quickly rule out the idea that the brown sacks of corn might represent some mysterious equipment related to missile launches.

“While satellite imagery can give you the keys to the castle for a country that is closed off, there is also lots of room for error that context,
” says David Schmerler
, a research associate at MIIS.

The MIIS researchers used near-infrared satellite imagery to highlight some rather large burn scars left in the surrounding vegetation from two North Korean missile launch attempts that both went awry in October 2016. The sequence of before and after daily images was taken by a constellation of smaller satellites owned by Planet. (More recently, the publication 
38 North
 also made use of Planet’s daily satellite imagery to examine landslides resulting from North Korea’s sixth nuclear test.)

Technical and cultural knowledge go hand-in-hand, says
 Joseph Bermudez Jr., CEO and co-founder of KPA Associates, LLC and an analyst focused on North Korea’s defense and intelligence affairs and third-world ballistic missile development. 
He recalled being sent an image where someone thought they had spotted filled-in bomb craters left over from the Korean War. In fact, the circles represented traditional North Korean burial mounds where a circular area is cleared of grass for a mound placed in the middle

“There is an incredible range to what people are saying about satellite images of North Korea, and the vast majority of it is inaccurate,” Bermudez Jr. says. 
“The reason for that is satellite imagery interpretation is a skill that needs to be learned or taught—it needs experience behind it before you reach a level of accuracy that is acceptable.”

Bermudez Jr. usually begins his work by using color correction, sharpening the image and other processing tricks to make a satellite image look as good as possible for inspection. He also rotates the image to the “look angle” that represents the direction from which the original image was captured.

If he has a general idea of where in the image is the activity of most interest—such as a missile launchpad—he will zoom in and compare it with previous images showing the same site. But he also will have his computer software pan slowly across the entire image while zoomed in so that he can get a good look at everything in high detail.

Diverse life experiences and real-world skills among a team of analysts can also prove incredibly helpful. Bermudez Jr. learned about structures as a fireman paramedic earlier in his life. As a wilderness instructor, he also spent a lot of time walking around in the mountains and understanding natural drainage systems such as rivers and valleys. Other a
nalysts may know more about
 
building ships or railroad operations. Each analyst can then build “interpretation keys” based on his or her knowledge to share with others in examining satellite images.

Satellite imagery analysts also benefit from leaning on outside experts in other fields. Hanham has previously tapped into the expertise of a cousin who is a long-haul trucker and her microbiologist mom’s knowledge of lab equipment. “I’ve leveraged my family,” she says.

The community of satellite imagery analysts with deep expertise on North Korea is small. That limited amount of human expertise may prove a bottleneck at a time when commercial satellites continue to build ever-larger databases of imagery on North Korea. But increased use of image-recognition algorithms based on machine learning could help, Hanham says. She and her MIIS colleagues hope to eventually have more of a semi-automated process of identifying certain objects.

Commercial satellites may soon provide even broader space surveillance of North Korea with a wider array of imaging equipment. Some already make use of hyperspectral imaging capabilities that go beyond the visible light spectrum. Others have begun mounting synthetic aperture radar (SAR) that can provide 3D mapping of the Earth’s surface regardless of weather or nighttime conditions.

But regardless of the technologies, interpretation of satellite imagery still depends heavily on the human expertise factor. And experienced analysts are always careful to caution that satellite images represent just once piece of a much larger puzzle when it comes to gathering intelligence on North Korea.

“I’ve run into many cases where people who are imagery analysts believe they have absolute truth,” Bermudez Jr. says. “Satellite imagery is just a single point in time that can only tell you so much.”


Reporting for this story was supported by the Stanley Foundation.

Powered by WPeMatico