AI for the Common Good
Stay or go? 18F and USDS insiders debate what it means to be in public service
On Nov. 9, 2016, Noah Kunin published a short Medium post explaining his decision to stay at 18F regardless of the results of the election. The rationale? The government startup was beyond politics. “My oath to this country was not to a particular office, or person, and certainly not to a political party. It was to the Constitution and to the people,” he wrote at the time.
Now, eight months later, Kunin has changed his tune. He’s leaving 18F.
Kunin’s personal evolution on the subject provides insight into how politics intrude into a nominally apolitical realm. The post-election tension was probably inevitable: Dozens of young liberal technologists were drawn early to 18F and to another Obama-era creation, the U.S. Digital Service and its corresponding agency arms.
‘The straw that broke my personal back’
For Kunin, three and a half years as an infrastructure director at 18F were brought to a halt by a line in former FBI Director James Comey’s testimony about President Donald Trump asking for Comey’s “loyalty.” This paired with the reorganization of GSA’s Technology Transformation Service under the Federal Acquisition Service — an organization controlled by a political appointee — led Kunin to worry that Trump might soon ask for loyalty from other public servants as well.
Loyalty to a specific leader, Kunin argues, undermines democracy. And though he hasn’t been asked for his loyalty directly, working with Trump’s administration means tacitly legitimizing its actions, he says.
Kunin’s argument stands in opposition to another popular opinion: The best way to work with the Trump administration is to show up and make divergent perspectives heard. That was the gist of the Washington Post op-ed Code For America founder Jennifer Pahlka published before her visit to the White House for Trump’s “tech week.”
It’s also an argument, Kunin says, that he has made before. “If the government is on fire, and you want to put that fire out,” he writes, “you have to run toward the fire. Not away from it.” But Kunin is no longer convinced this is an ordinary fire situation.
Think about a grease fire. Putting out a fire like that doesn’t work by running in and throwing water on it. That only spreads the fire around. To put out an oil fire, you have to suffocate it of the oxygen it needs to burn. In our democracy, that oxygen is “consent to power”. Individuals, corporations, organizations need to think hard and long about this, before working with or for this Administration in any capacity.
A philosophical question
Kunin told Business Insider that the aim of his Medium post, besides revealing his own decision, is to remind his fellow public servants to evaluate their position.
“I don’t think that would be a fair take that it’s a broad call for mass resignation from the public service,” Kunin said. “The post is a call for public servants to ask the question of complicity of themselves, and the nature of their work. This is not a ‘one size fits all’ solution.”
Kunin’s tenure at 18F — about the length of one presidential term — is fairly typical for people drawn to doing a “tour of duty” in public service. Most 18F employees serve in “not to exceed” positions which means they serve for a term of two years, extendable to a total of four years. As such, various 18F and USDS officials departed recently with this kind of length of service on their resumes.
They generally left quietly. Kunin, instead, chose to use the opportunity to start a conversation, and there was an audience for it. Vivian Graubard, a founding member of USDS who is now at the New America think tank, tweeted her agreement with his perspective:
https://twitter.com/vivigraubard/status/882679462504140800
Chris Cairns, a cofounder of 18F now out in the private sector, also understands where Kunin is coming from. “I share his concern about the potential politicization of 18F, regardless of administration,” he wrote in an email to FedScoop. “It only makes sense to be in the room so long as your voice is being heard.” Cairns left last month after about four years at 18F.
Chris Given, who works for USDS at the Department of Veterans Affairs, responded to Kunin’s post with a series of tweets explaining why he’s still on the inside. “I stay because 1) I know my work improves people’s lives, and 2) I have not been asked, directly or in kind, for my personal loyalty,” he tweeted. That said, he wrote, he spends time thinking about how he’d respond to an unethical ask. “My eyes are open. I start days knowing that I might have to quit before their end. I don’t relish that possibility, but I am ready.”
Aaron Snow, another 18F cofounder, is concerned about throwing the baby out with the bathwater. “I think being there [at 18F] remains important,” he said. “The federal government is a vast enterprise and there are many ways to provide public service in its employ, regardless of whether the president agrees with you.”
Snow, who also finished a four-year stretch and is no longer part of the government, reiterated his respect for Kunin and others who may decide to leave public service early. But being able to leave a job based on a philosophical difference of opinion is a position of privilege, he cautioned. The federal government employs millions of Americans — people who continue dispersing Social Security checks and implementing tax policy and maintaining federal buildings regardless of the president in charge. It’s how the government works.
Should 18F employees be any different? At the end of the day these decisions are personal, and based on individual circumstance, Snow said.
“Everybody,” he said, “has to draw their own lines.”
How We’re Bringing AI to the Fight Against Cancer
Cancer is the epidemic of our time – over half a million Americans die each year from cancer – more than 1,500 each day. 15 million of our family members, friends, and colleagues are currently living with this disease.
This is why we were excited when the White House announced the Cancer Moonshot program earlier year, declaring a goal to make a decade worth of advances in cancer prevention and treatment in just 5 years.
NVIDIA has teamed up with the National Cancer Institute, the U.S. Department of Energy, and national research laboratories to build a common discovery platform for cancer called CANDLE, based on today’s state of the art technology artificial intelligence (AI).
The field of AI has made huge leaps in the last five years. Using Deep learning — a technique where computers teach themselves from massive volumes of data — we have already achieved superhuman results. Take, for example, the field of speech recognition: researchers for the last 20 years had achieved an accuracy rate of only 70%, which was considered unusable. With deep learning on NVIDIA GPUs, Microsoft and Baidu have already surpassed human capability in speech recognition in just 3 years.
We now know that AI required three key elements to be practical: massive amounts of data, sophisticated algorithms and high-performance parallel processors (GPUs). We will apply those same three ingredients, feeding petabytes of cancer data from the NCI into CANDLE, a deep learning platform for the nation’s most advanced GPU-accelerated supercomputers.
NVIDIA engineers and computational scientists will contribute to this platform by developing an AI software framework optimized for the latest supercomputing infrastructure, with the goal of achieving 10X annual increases in productivity for cancer researchers.
The CANDLE development is targeted at three precision medicine pilot projects. Deep learning techniques are essential to each.
- CANDLE will help researchers discover the underlying genetic signatures in the DNA and RNA of common cancers. These signatures can help researchers tap into molecular data collected by the Cancer Moonshot Initiative to predict how patients will respond to treatment.
- CANDLE will accelerate the molecular dynamic simulations of key protein interactions. This will help researchers understand the underlying biological mechanisms creating conditions for cancer.
- CANDLE will automate information extraction and analysis of millions of clinical patient records to build a comprehensive cancer surveillance database of disease metastasis and recurrence.
Complicated stuff. But the goal of all these efforts is simple: expedite individual treatments, discover new treatments faster, and more accurately predict how each patient’s cancer will evolve.
These are hard challenges. They’ll require great determination and coordination among many researchers and scientists. By using deep learning we hope to achieve the goals of the Cancer Moonshot, making giant leap for mankind.
To learn more, we will be continuing the discussion on the battle against cancer with Jerry Lee, Health Science Director at NCI and Cancer Moonshot Taskforce member, at GTC DC 2017.
The Era of AI Computing
At GTC, we unveiled Volta, our greatest generational leap since the invention of CUDA. It incorporates 21 billion transistors. It’s built on a 12nm NVIDIA-optimized TSMC process. It includes the fastest HBM memories from Samsung. Volta features a new numeric format and CUDA instruction that perform 4×4 matrix operations–an elemental deep learning operation–at super-high speeds.
Each Volta GPU is 120 teraflops. And our DGX-1 AI supercomputer interconnects eight Tesla V100 GPUs to generate nearly one petaflops of deep learning performance.
Google’s TPU
Also last week, Google announced at its I/O conference, its TPU2 chip, with 45 teraflops of performance.
It’s great to see the two leading teams in AI computing race while we collaborate deeply across the board–tuning TensorFlow performance, and accelerating the Google cloud with NVIDIA CUDA GPUs. AI is the greatest technology force in human history. Efforts to democratize AI and enable its rapid adoption are great to see.
Powering Through the End of Moore’s Law
As Moore’s law slows down, GPU computing performance, powered by improvements in everything from silicon to software, surges.
The AI revolution has arrived despite the fact Moore’s law–the combined effect of Dennard scaling and CPU architecture advance–began slowing nearly a decade ago. Dennard scaling, whereby reducing transistor size and voltage allowed designers to increase transistor density and speed while maintaining power density, is now limited by device physics.
CPU architects can harvest only modest ILP–instruction-level parallelism–but with large increases in circuitry and energy. So, in the post-Moore’s law era, a large increase in CPU transistors and energy results in a small increase in application performance. Performance recently has increased by only 10 percent a year, versus 50 percent a year in the past.
The accelerated computing approach we pioneered targets specific domains of algorithms; adds a specialized processor to offload the CPU; and engages developers in each industry to accelerate their application by optimizing for our architecture. We work across the entire stack of algorithms, solvers and applications to eliminate all bottlenecks and achieve the speed of light.
That’s why Volta unleashes incredible speedups for AI workloads. It provides a 5X improvement over Pascal, the current-generation NVIDIA GPU architecture, in peak teraflops, and 15X over the Maxwell architecture, launched just two years ago-–well beyond what Moore’s law would have predicted.
Accelerate Every Approach to AI
A sprawling ecosystem has grown up around the AI revolution.
Such leaps in performance have drawn innovators from every industry, with the number of startups building GPU-driven AI services growing more than 4x over the past year to 1,300.
No one wants to miss the next breakthrough. Software is eating the world, as Marc Andreessen said, but AI is eating software.
The number of software developers following the leading AI frameworks on the GitHub open-source software repository has grown to more than 75,000 from fewer than 5,000 over the past two years.
The latest frameworks can harness the performance of Volta to deliver dramatically faster training times and higher multi-node training performance.
Deep learning is a strategic imperative for every major tech company. It increasingly permeates every aspect of work from infrastructure and tools to how products are made. We partner with every framework maker to wring out the last drop of performance. By optimizing each framework for our GPU, we can improve engineer productivity by hours and days for each of the hundreds of iterations needed to train a model. Every framework–Caffe2, Chainer, Microsoft Cognitive Toolkit, MXNet, PyTorch, TensorFlow–will be meticulously optimized for Volta.
The NVIDIA GPU Cloud platform gives AI developers access to our comprehensive deep learning software stack wherever they want it—on PCs, in the data center or via the cloud.
We want to create an environment that lets developers do their work anywhere, and with any framework. For companies that want to keep their data in-house, we introduced powerful new workstations and servers at GTC.
Perhaps the most vibrant environment is the $247 billion market for public cloud services. Alibaba, Amazon, Baidu, Facebook, Google, IBM, Microsoft and Tencent all use NVIDIA GPUs in their data centers.
To help innovators move seamlessly to cloud services such as these, at GTC we launched the NVIDIA GPU Cloud platform, which contains a registry of pre-configured and optimized stacks of every framework. Each layer of software and all of the combinations have been tuned, tested and packaged up into an NVDocker container. We will continuously enhance and maintain it. We fix every bug that comes up. It all just works.
A Cambrian Explosion of Autonomous Machines
Deep learning’s ability to detect features from raw data has created the conditions for a Cambrian explosion of autonomous machines–IoT with AI. There will be billions, perhaps trillions, of devices powered by AI.
At GTC, we announced that one of the 10 largest companies in the world, and one of the most admired, Toyota, has selected NVIDIA for their autonomous car.
We also announced Isaac, a virtual robot that helps make robots. Today’s robots are hand programmed, and do exactly and only what they were programmed to do. Just as convolutional neural networks gave us the computer vision breakthrough needed to tackle self-driving cars, reinforcement learning and imitation learning may be the breakthroughs we need to tackle robotics.
Once trained, the brain of the robot would be downloaded into Jetson, our AI supercomputer in a module. The robot would stand, adapt to any differences between the virtual and real world. A new robot is born. For GTC, Isaac learned how to play hockey and golf.
Finally, we’re open-sourcing the DLA, Deep Learning Accelerator–our version of a dedicated inferencing TPU–designed into our Xavier superchip for AI cars. We want to see the fastest possible adoption of AI everywhere. No one else needs to invest in building an inferencing TPU. We have one for free–designed by some of the best chip designers in the world.
Enabling the Einsteins and Da Vincis of Our Era
These are just the latest examples of how NVIDIA GPU computing has become the essential tool of the da Vincis and Einsteins of our time. For them, we’ve built the equivalent of a time machine. Building on the insatiable technology demand of 3D graphics and market scale of gaming, NVIDIA has evolved the GPU into the computer brain that has opened a floodgate of innovation at the exciting intersection of virtual reality and artificial intelligence.
Learn the latest from NVIDIA on AI and Deep Learning in our newsletter.
This Drone Goes Where GPS Can’t
Most drones would be lost without GPS. Not this one.A drone developed by NVIDIA researchers navigates even the most far-flung, unmapped places using only deep learning and computer vision powered by NVIDIA Jetson TX1 embedded AI supercomputers.
Although initially designed to follow forest trails to rescue lost hikers or spot fallen trees, the low-flying autonomous drone could work far beyond the forest — in canyons between skyscrapers or inside buildings, for example — where GPS is inaccurate or unavailable.
“This works when GPS doesn’t,” said Nikolai Smolyanskiy, the NVIDIA team’s technical lead. “All you need is a path the drone can recognize visually.”
Researchers built their drone with off-the-shelf components to reduce costs.
No GPS? No Problem
Although the technology is still experimental, it could eventually search for survivors in damaged buildings, inspect railroad tracks in tunnels, check stock on store shelves, or adapted to examine communications cables underwater, Smolyanskiy said.
The team’s already trained it to follow train tracks and ported the system to a robot-on-wheels to traverse hallways. The drone also avoids obstacles like people, pets or poles.
“We chose forests as a proving ground because they’re possibly the most difficult places to navigate,” he said. “We figured if we could use deep learning to navigate in that environment, we could navigate anywhere.”
Unlike a more urban environment, where there’s generally uniformity to, for example, the height of curbs, shape of mailboxes and width of sidewalks, the forest is relatively chaotic. Trails in the woods often contain no markings. Light can be filtered through leaves; it also varies from bright sunlight to dark shadows. And trees vary in height, width, angle and branches.
Flight Record
To keep costs low, the researchers built their device using an off-the-shelf drone equipped with the NVIDIA Jetson TX1 and two cameras.
“Our whole idea is to use cameras to understand and navigate the environment,” Smolyanskiy said. “Jetson gives us the computing power to do advanced AI onboard the drone, which is a requirement for operating in remote environments.”
The NVIDIA team isn’t the first to pursue a drone that navigates without GPS, but the researchers achieved what they believe is the longest and most stable flight of its kind. Their fully autonomous drone flies along the trail for a kilometer (about six-tenths of a mile), avoiding obstacles and maintaining a steady position in the center of the trail.
Team member Alexey Kamenev played a big role in making this happen. He developed deep learning techniques that allowed the drone to smoothly fly along trails without sudden movements that would make it wobble. He also reduced the need for massive amounts of data typically needed to train a deep learning system.
In the video below, the drone follows a trail in the forest near the researchers’ Redmond, Wash., office. The areas in green are where the robot decided to fly and the red areas are those it rejected.
No Breadcrumbs Needed
The drone learned to find its way by watching video that Smolyanskiy shot along eight miles of trails in the Pacific Northwest. He took the video in different lighting conditions with three wide-angle GoPro cameras mounted on the left, center and right of a metal bar on a mini Segway.
In addition to their own footage, researchers trained their neural network — called TrailNet — on video recorded on trails in the Swiss Alps by AI researchers at Istituto Dalle Molle di Studi sull’Intelligenza Artificiale (IDSIA) in Zurich.
In fact, IDSIA’s work on drone forest navigation was one inspiration for NVIDIA’s autonomous drone team. The other inspiration was NVIDIA’s self-driving car, BB8.
Next Steps
The team now plans to create downloadable software for Jetson TX1 and Jetson TX2 so others can build robots that navigate based on visual information alone.
Long term, the idea is to tell the robot to travel between two points on any map — whether it’s a Google map or a building plan — and have it successfully make the trip, avoiding obstacles along the way.
To hear the latest talks on drone development and its impact upon our daily lives, register for GTC DC 2017.
Steve Rice, former TSA CIO, is now principal deputy CIO at DHS
Steve Rice is now the principal deputy chief information officer at the Department of Homeland Security. He took the position in January and has been overseeing the “management, security, and sharing of DHS information technology” since then, the agency told FedScoop.
Rice comes directly from the Transportation Security Administration, where he served as CIO and assistant administrator for the Office of Information Technology. He took that role in 2013, having previously worked as deputy CIO. While at TSA, Rice oversaw IT strategy, operations and security.
At DHS Rice works under CIO Richard Staropoli, who was appointed by President Trump at the end of April. Both men share a professional history within the Secret Service — Staropoli served as a special agent for more than two decades. Rice, meanwhile, worked in both intelligence and tech roles in the 1990s, assisting the unit that responded to the Oklahoma City bombings, for example.
Military testing behavioral ID technology that would replace CAC card
The Pentagon has finally inked a deal to pilot behavioral biometric technology to identify those using its computer networks, more than a year after then-CIO Terry Halvorsen first pledged to get rid of the ubiquitous Common Access Card.
Shaun Waterman of CyberScoop reports that Vancouver, Canada-based Plurilock will provide the technology. The company’s BioTrack product develops a unique profile of users based on the way they interact with computer keyboards, mice and touchscreens.
“Plurilock’s advanced system for determining ongoing proof of presence provides a cybersecurity solution that instantaneously recognizes breaches, helps with corporate forensic investigations, and ensures regulatory compliance.” said Plurilock CEO Ian Paterson.
As the MGT Act faces the Senate, are you ready for what’s next?
Leaders in Congress, the administration and across industry are all on the same page: Federal IT is in need of a historic transformation and we are going to work together to get it done. For years we watched as lack of funds or bureaucratic roadblocks limited the government’s ability to provide modern citizen services. The Modernizing Government Technology (MGT) Act was recently passed in the House with overwhelming bipartisan support and a companion bill is currently making its way through the Senate. MGT can help agencies modernize their IT and application services, but CIOs must prepare.
If recently passed Federal IT Acquisition Reform Act legislation is the stick that Congress is using to push agencies toward modernization, MGT is the first carrot Congress is willing to give out to bring agencies up to speed. In its current version, the MGT Act calls for a central fund to support IT transformation – providing seed money for CIOs to move a modernization project off the wish list and into action. In a show of support for the bill, the Trump administration’s fiscal 2018 budget proposal establishes the modernization fund outlined in the MGT Act.
With new funding streams on the horizon, now is the time to prepare. Here are a few suggestions to help you get ready:
Inventory
As the first step to fast-track modernization, federal CIOs need to inventory agency IT assets. Because of FITARA, CIOs are in a better position today to drive change, but without a complete inventory of their technology investments, it is impossible for them to make informed decisions on new investments. An effective modernization strategy starts by identifying what needs to be accomplished first.
Identify
Knowing your asset portfolio, and the business goals it serves, will guide you to which projects are ripe for investment and which ones should wait. MGT offers a relatively modest amount of money for agencies, so CIOs should try to find a proof-of-concept project that can make use of the money and also serve as a quick win on which future modernization efforts can be built. However, the identification process is key, not only to utilizing MGT funds, but also truly understanding the full spectrum of technology solutions that your agency needs to run as an effective and modern organization.
Partner
In early June, the Government Accountability Office announced its fourth iteration of FITARA scorecards, awarding the first “A” in the history of the program to the U.S. Agency for International Development. The agency had routinely received “Ds” on previous scorecards. Rep. Gerry Connelly, D-Va., praised USAID for asking the GAO’s advice on how to improve its scores. Additionally, agencies can leverage expertise from the General Services Administration and other agency CIOs, as well as take advantage of industry expertise in IT transformation.
Transform
Wherever you choose to begin IT modernization, moving data and applications to the cloud awards you tremendous flexibility. A hybrid cloud environment can increase efficiency, security and cost savings. A hybrid approach also allows agencies to run a variety of systems and applications within their technology stack while maintaining greater control over their data.
Software-defined data centers offer agencies simplicity and flexibility in managing upgrades across the entire system from a single point. SDDCs improve automation, increase agility and decrease the potential for manual error. Automated, flexible environments allow agencies to meet user demand based on tailored requirements, reducing IT overhead costs.
While MGT is not yet final, CIOs should have a plan in place for new funding streams. Inventory assets; identify the best projects to use as a proof-of-concept; utilize resources from across government and industry. By taking these steps, agencies can begin to transform their legacy infrastructure and provide modern technology services to their users.
Steve Harris is vice president and general manager of federal for Dell EMC.
White House Historical Association dives into more digitization
The White House Historical Association is all about connecting Americans with 1600 Pennsylvania Avenue’s illustrious past, and the organization recently expanded a relationship with Amazon Web Services to make more resources available through modern technology.
The nonprofit group’s Digital Library, which is hosted on AWS, is collection of historical images of the White House, including interior shots, exterior photos, decorative artwork and more. The association recently celebrated the library’s one-year launch anniversary, and announced that the collection now contains over 5,000 images.
The association wants to do more, and in June it unveiled a new initiative with AWS that will “allow significant expansion of the Library.”
“We’ve partnered with Amazon Web Services to rethink and expand the architecture of our cloud storage to allow for more workflows to occur simultaneously,” the Historical Association told FedScoop. Practically this means the Association is now using Amazon Snowball to upload a backlog of around seven terabytes of data to Fotoware Digital Asset Management, the software that runs the Digital Library. This is crucial, the association said, because “it creates a secure cloud backup of what were files that lived only on local drives and were at risk.”
The association, which receives much of its funding through private donations, is also working on digitizing 25,000 previously-hidden 35mm slides taken between 1960 and 1990, a spokesperson said.
In addition to digitizing and sharing more artifacts, the association announced plans to create a mobile app to provide virtual tours of the White House. According to a press release, the app will “focus on major historical elements in famous rooms of the White House, allowing people across the country to experience a White House tour.” A spokesperson for the association declined to comment further on the app, saying only that it is in very early stages.
“An important part of the White House Historical Association’s mission is education – especially for those that can’t experience it in Washington,” Stewart D. McLaurin, the group’s president, said in a statement. “By leveraging AWS, we will increase public access and bring White House history across the country to millions of Americans.”
In January the organization also launched a podcast called 1600 Sessions, which seeks to tell “the stories and traditions of the executive mansion where the president works and the first family lives.”
DHS needs better information security practices, audit says
The Department of Homeland Security needs to up its game on information security, according to an audit released last week.
Private sector auditor KPMG conducted after-hours walkthroughs of employee workstations in the department’s Office of Financial Management and the Office of the Chief Information Officer, and found sensitive information — like passwords — left out and unattended.
Auditors also found unsecured government-issued laptops and mobile devices. Of the 69 workstations KPMG inspected, three breached DHS information security policy.
The audit, conducted during fiscal 2016, also reviewed DHS financial statements and found that both the OFM and OCIO use password configurations that don’t meet agency standards.
While KPMG found the physical security behavior certainly needs improvement, it did take care to note that the three unsecured workstations don’t necessarily reveal a workplace-wide trend. “The selection of inspected areas was not statistically derived; therefore, the results described here should not be used to extrapolate to OFM and OCIO as a whole,” the report says.
A separate KPMG audit, also released last week, surveyed DHS’ National Protection and Programs Directorate, and found similar information security weaknesses. For example, the report notes, account management policies at the directorate are too vague.
“Account management policies did not exist or were lacking sufficient detail in areas such as segregation of duties, recertification, elevated privileges, and disabling accounts upon user separation,” the audit states.
According to the auditors, this issue and others “collectively limited NPPD’s ability to ensure that critical financial and operational data were maintained in such a manner as to ensure their confidentiality, integrity, and availability.”