Category: ITManagementNews

ITManagementNews

Over 1 Billion YouTube Videos Now Have Captions

“I envision a future where everything will be captioned, so the more than 300 million people who are deaf or hard of hearing like me will be able to enjoy videos like everyone else,” said Liat Kaver, a YouTube Product Manager focusing on captions and accessibility. “When I was growing up in Costa Rica, there were no closed captions in my first language, and only English movies had Spanish subtitles. I felt I was missing out because I often had to guess at what was happening on the screen or make up my own version of the story in my head. That was where the dream of a system that could just automatically generate high quality captions for any video was born.”

Google first launched a video captions option in 2006 followed by automated captions in 2009. Captions are now supported in 10 languages.

As YouTube grew, so did the number of videos with captions which now stands at over 1 billion. Kaver says that more than 15 million videos are watched each day with captions enabled.

Caption Technology

“One of the ways that we were able to scale the availability of captions was by combining Google’s automatic speech recognition (ASR) technology with the YouTube caption system to offer automatic captions for videos,” says Kaver. “There were limitations with the technology that underscored the need to improve the captions themselves. Results were sometimes less than perfect, prompting some creators to have a little fun at our expense!”

Kaver says that one of their teams major goals has been to improve automatic caption accuracy via technological improvements in speech recognition, machine learning and increases in training data. “All together, those technological efforts have resulted in a 50 percent leap in accuracy for automatic captions in English, which is getting us closer and closer to human transcription error rates,” she says. “I know from firsthand experience that if you build with accessibility as a guiding force, you make technology work for everyone.”

February 20, 2017
Apple Publishes First AI Research Paper on Using Adversarial Training to Improve Realism of Synthetic Imagery

Earlier this month Apple pledged to start publicly releasing its research on artificial intelligence. During the holiday week, Apple has released its first AI research paper detailing how its engineers and computer scientists used adversarial training to improve the typically poor quality of synthetic, computer game style images, which are frequently used to help machines learn.

The paper’s authors are Ashish Shrivastava, a researcher in deep learning, Tomas Pfister, another deep learning scientist at Apple, Wenda Wang, Apple R&D engineer, Russ Webb, a Senior Research Engineer, Oncel Tuzel, Machine Learning Researcher and Joshua Susskind, who co-founded Emotient in 2012 and is a deep learning scientist.

The team describes their work on improving synthetic images to improve overall machine learning:

With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations. However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions. To reduce this gap, we propose Simulated+Unsupervised (S+U) learning, where the task is to learn a model to improve the realism of a simulator’s output using unlabeled real data, while preserving the annotation information from the simulator.

We developed a method for S+U learning that uses an adversarial network similar to Generative Adversarial Networks (GANs), but with synthetic images as inputs instead of random vectors. We make several key modifications to the standard GAN algorithm to preserve annotations, avoid artifacts and stabilize training: (i) a ‘self-regularization’ term, (ii) a local adversarial loss, and (iii) updating the discriminator using a history of refined images. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study.

We quantitatively evaluate the generated images by training models for gaze estimation and hand pose estimation. We show a significant improvement over using synthetic images, and achieve state-of-the-art results on the MPIIGaze dataset without any labeled real data.

Conclusions and Future Work

“We have proposed Simulated+Unsupervised learning to refine a simulator’s output with unlabeled real data,” says the Apple AI Scientists. “S+U learning adds realism to the simulator and preserves the global structure and the annotations of the synthetic images. We described SimGAN, our method for S+U learning, that uses an adversarial network and demonstrated state-of-the-art results without any labeled real data.”

They added, “In future, we intend to explore modeling the noise distribution to generate more than one refined image for each synthetic image, and investigate refining videos rather than single images.”

View the research paper (PDF).

February 13, 2017
IBM Watson Brings AI to H&R Block Tax Preparation

IBM announced a partnership with H&R block to use their artificial intelligent platform IBM Watson to radically improve tax preparation. “Introducing the biggest advancement in tax preparation technology,” exclaimed IBM in an announcement video. “Say hello to the partnership between H&R Block and IBM Watson. Imagine being able to understand all 74,000 pages of the US tax code along with thousands of yearly tax law changes and other information, plus locks deep insights built from over 600 million data points.”

“Imagine being able to understand all that information,” noted IBM. “Watson will learn from it and help your tax pro find every credit, deduction and opportunity available. The one of a kind partnership between H&R Block and Watson is revolutionizing the way people file taxes.”

H&R Block is marketing the new AI integration as “the future of tax prep” as seen in their new Google ad:

“H&R Block is revolutionizing the tax filing experience,” stated Bill Cobb, President and Chief Executive Officer of H&R Block. “By combining the human expertise, knowledge and judgement of our tax pros with the cutting edge cognitive computing power of Watson, we are creating a future where every last deduction and credit can be found.”

“Tax preparation is a perfect use for Watson,” noted David Kenny, Senior Vice President of IBM Watson. “Just like Watson is already revolutionizing other industries like healthcare and education, here H&R Block with Watson is learning to process incredible amounts of information, helping create tailored solutions for H&R Block customers.”

IBM expects Watson to learn through H&R Blocks millions of unique tax filings how to maximize credits and deductions for every customer, elimination inconsistencies caused by human tax preparation experts. The more information Watson receives says Kenny, the smarter Watson gets.

“This is a major shift in how man and machine work together to help us in our everyday lives,” says Kenny.

February 6, 2017
Microsoft CEO: We Are Not Anywhere Close To Achieving Artificial General Intelligence

Satya Nadella, CEO of Microsoft, recently was interviewed by Ludwig Siegele of The Economist about the future of AI (artificial intelligence) at the DLD in Munich, Germany where he spoke about the need to democratize the technology so that it is part of every company and every product. Here’s an excerpt transcribed from the video interview:

What is AI?

The way I have defined AI in simple terms is we are trying to teach machines to learn so that they can do things that humans do, but in turn help humans. It’s augmenting what we have. We’re still in the mainframe era of it.

There has definitely been an amazing renaissance of AI and machine learning. In the last five years there’s one particular type of AI called deep neural net that has really helped us, especially with perception, our ability to hear or see. That’s all phenomenal, but if you ask are we anywhere close to what people reference, artificial general intelligence… No. The ability to do a lot of interesting things with AI, absolutely.

The next phase to me is how can we democratize this access? Instead of worshiping the 4, 5 or 6 companies that have a lot of AI, to actually saying that AI is everywhere in all the companies we work with, every interface, every human interaction is AI powered.

What is the current state of AI?

If you’re modeling the world, or actually simulating the world, that’s the current state of machine learning and AI. But if you can simulate the brain and the judgements it can make and transfer learning it can exhibit… If you can go from topic to topic, from domain to domain and learn, then you will get to AGI, or artificial general intelligence. You could say we are on our march toward that.

The fact that we are in those early stages where we are at least being able to recognize and free text, things like keeping track of things, by modeling essentially what it knows about me and my world and my work is the stage we are at.

Explain democratization of AI?

Sure, 100 years from now, 50 years from now, we’ll look back at this era and say there’s been some new moral philosopher who really set the stage as to how we should make those decisions. In lieu of that though one thing that we’re doing is to say that we are creating AI in our products, we are making a set of design decisions and just like with the user interface, let’s establish a set of guidelines for tasteful AI.

The first one is, let’s build AI that augments human capability. Let us create AI that helps create more trust in technology because of security and privacy considerations. Let us create transparency in this black box. It’s a very hard technical problem, but let’s strive toward saying how do I open up the black box for inspection?

How do we create algorithm accountability? That’s another very hard problem because I can say I created an algorithm that learns on its own so how can I be held accountable? In reality we are. How do we make sure that no unconscious bias that the designer has is somehow making it in? Those are hard challenges that we are going to go tackle along with AI creation.

Just like quality, in the past we’ve thought about security, quality and software engineering. I think one of the things we find is that for all of our progress with AI the quality of the software stack, to be able to ensure the things we have historically ensured in software are actually pretty weak. We have to go work on that.

January 30, 2017
Google Creates a Technical Guide for Moving to the Cloud

Google has created a guide in the form of a website for companies that are considering a move to their cloud called Google Cloud Platform for Data Center Professionals.

“We recognize that a migration of any size can be a challenging project, so today we’re happy to announce the first part of a new resource to help our customers as they migrate,” said Peter-Mark Verwoerd,a Solutions Architect at Google who previously worked for Amazon Web Services. “This is a guide for customers who are looking to move to Google Cloud Platform (GCP) and are coming from non-cloud environments.”

The guide focuses on the basics of running IT — Compute, Networking, Storage, and Management. “We’ve tried to write this from the point of view of someone with minimal cloud experience, so we hope you find this guide a useful starting point,” said Verwoerd.

January 30, 2017
Google Using RAISR Technology on Google+ and Saving 75% in Bandwidth

Google+ has become a haven for high end photos by professional photographers who obviously care about image quality. Google’s solution to the huge bandwidth requirements for their free service is a technology called RAISR. Lower bandwidth is also a benefit to the end user by increasing loading speeds and lowering data costs. This is especially concerning outside of the United States where it’s rare not to have to pay for internet based on data usage.

Back in November Google introduced a machine learning technology called “RAISR: Rapid and Accurate Image Super-Resolution”, that creates high-quality versions of low-resolution images. “RAISR produces results that are comparable to or better than the currently available super-resolution methods, and does so roughly 10 to 100 times faster, allowing it to be run on a typical mobile device in real-time,” explained Peyman Milanfar, Lead Scientist at Google Research. “Furthermore, our technique is able to avoid recreating the aliasing artifacts that may exist in the lower resolution image.”

Here’s how Google’s technical team (Yaniv Romano, John Isidoro, Peyman Milanfar) described it in June 2016:

Given an image, we wish to produce an image of larger size with significantly more pixels and higher image quality. This is generally known as the Single Image Super-Resolution (SISR) problem. The idea is that with sufficient training data (corresponding pairs of low and high resolution images) we can learn set of filters (i.e. a mapping) that when applied to given image that is not in the training set, will produce a higher resolution version of it, where the learning is preferably low complexity. In our proposed approach, the run-time is more than one to two orders of magnitude faster than the best competing methods currently available, while producing results comparable or better than state-of-the-art.

A closely related topic is image sharpening and contrast enhancement, i.e., improving the visual quality of a blurry image by amplifying the underlying details (a wide range of frequencies). Our approach additionally includes an extremely efficient way to produce an image that is significantly sharper than the input blurry one, without introducing artifacts such as halos and noise amplification. We illustrate how this effective sharpening algorithm, in addition to being of independent interest, can be used as a pre-processing step to induce the learning of more effective upscaling filters with built-in sharpening and contrast enhancement effect.

“RAISR, which was introduced in November, uses machine learning to produce great quality versions of low-resolution images, allowing you to see beautiful photos as the photographers intended them to be seen,” noted John Nack, Product Manager of Digital Photography at Google. “By using RAISR to display some of the large images on Google+, we’ve been able to use up to 75 percent less bandwidth per image we’ve applied it to.”

“While we’ve only begun to roll this out for high-resolution images when they appear in the streams of a subset of Android devices, we’re already applying RAISR to more than 1 billion images per week, reducing these users’ total bandwidth by about a third,” said Nack. “In the coming weeks we plan to roll this technology out more broadly — and we’re excited to see what further time and data savings we can offer.”

January 30, 2017
Government Can Speed Up Implementation of IoT Technology
Government around the world play a key role in whether IoT becomes a mainstream technology sooner rather than later according to Cisco IoT expert Maciej Kranz. Kranz recently posted an excerpt of his book Building the Internet of Things on the Cisco Innovation blog.

IoT Adoption is Key to Regional Competitiveness

“Governments around the world are beginning to realize that IoT adoption will be one of the key factors defining the competitiveness of their cities, provinces, countries, or regions and that IoT can help solve many of the chronic problems plaguing their economies and their environments,” says Kranz. “Thus, governments at various levels have a number of key roles to play.”

“There will be competition for bandwidth and other resources; there will be ideas that may conflict with public policy; and there will be IoT-based ideas that need to be regulated to ensure public safety and privacy,” noted Kranz. “Think drones. In these and other ways, government regulations can help direct and align the industry.”

Kranz offered a few examples of U.S. legislations and related impact:
- The Energy Act drove the need for energy monitoring, including smart meters.
- The Rail Safety Improvement Act specified the requirements and the deadline (since extended) for adoption of Positive Train Control on main U.S. railways.
- The Food Safety Modernization Act drove the requirements for IoT-based systems, including quality control and source tracking, across the food supply chain to prevent food safety issues.
- Most recently, the Drug Quality and Security Act requires the adoption of a system to identify and trace prescription drugs.
Kranz believes that government funding priorities may drive the future of IoT. “Through their spending power, governments can drive the focus and accelerate the adoption of IoT technologies and solutions. In aggregate, governments represent a huge global market. Their priorities, what they choose to buy, and what problems they choose to address can drive the roadmaps of IoT technology and solution providers.”

He lists these additional government roles:
- Supporting training and education
- Supporting development of startup ecosystems
- Supporting standards efforts
- Supporting basic research and development
- Enabling competitiveness and openness of the country’s markets
- Promoting best practices and modern business models
Why the IoT is Important to Our Future

Kranz’ promo video for his book says this about the amazing future predicted for IoT technology, impacting not just consumers but manufacturers and really… everybody.

“The wheel, printing press, the airplane. It’s impossible to imagine life without them and soon it will be just as impossible to imagine life before the Internet of Things! IoT is already happening and the growth and opportunity it provides isn’t just big, it’s huge. Wheel, printing press and airplane huge. Billions of connected devices, trillions in revenue.”

At its core, Kranz said on his website, “it’s about business outcomes and people; it is about new ways of doing business, talent and change management; it is about migration to open technologies and open business structures based on co-development and ecosystems of partnerships; it is a multi-year, multi-phase journey.”

Here’s a recent interview that Maciej Kranz gave explaining IoT to investors:
January 29, 2017
Google Code-in for Teens is Wildly Successful
Google is in the middle of its annual Code-In contest and it’s more popular than ever with 930 teenagers from 60 countries completing 3,503 tasks with 17 open source organizations. “The number of students successfully completing tasks has almost met the total number of students from the 2015 contest already,” said Stephanie Taylor, Google Code-in Program Manager in a blog post on the Google Open Source Blog. This is the 7th year of Google Code-in.

Tasks that the students have completed include:
- writing test suites
- improving mobile UI
- writing documentation and creating videos to help new users
- working on internationalization efforts
- fixing and finding bugs in the organization’s’ software
Check it out: Google Code-in Website

What is Google Code-in?

Google Code-in is a way for Google to inspire young students to enter the field of software development. “Don’t wait until they are university students,” said Taylor in a talk at GSoC (Google Summer of Code) earlier this year in Singapore. “Let’s get them excited about open source when they are 13, 14, 15 years old. So Google Code-in was born.”

Google Code-in is an online global contest for 13-17 year old students around the world.

There’s Still Time to Get Started!

“Students, there is still plenty of time to get started with Google Code-in,” said Taylor. “New tasks are being added daily to the contest site — there are over 1,500 tasks available for students to choose from right now! If you don’t see something that interests you today, check back again every couple of days for new tasks.”

She says that the last day to register for the contest and claim a task is Friday, January 13, 2017 with all work being due on Monday, January 16, 2017 at 9:00 am PT.
December 27, 2016
Twas the Night Before Christmas and All Through the Cloud…

The Google Cloud Team posted a fun poem for all of us techno nerds. “2016 is winding down, and we wanted to take this chance to thank you, our loyal readers, and wish you happy holidays,” wrote Alex Barrett, Editor of the Google Cloud Platform Blog. “As a little gift to you, here’s a poem, courtesy of Mary Koes, a product manager on the Stackdriver team channeling the Clement Clarke Moore classic.”

Twas the night before Christmas and all through the Cloud
Not a creature was deploying; it wasn’t allowed.
The servers were all hosted in GCP or AWS
And Stackdriver was monitoring them so no one was stressed.

The engineers were nestled all snug in their beds
While visions of dashboards danced in their heads.
When then from my nightstand, there arose such a clatter,
I silenced my phone and checked what was the matter.

Elevated error rates and latency through the roof?
At this rate our error budget soon would go poof!
The Director OOO, the CTO on vacation,
Who would I find still manning their workstation?

Dutifully, I opened the incident channel on Slack
And couldn’t believe when someone answered back.
SClaus was the user name of this tireless engineer.
I wasn’t aware that this guy even worked here.

He wrote, “Wait while I check your Stackdriver yule Logs . . .
Yep, it seems the errors are all coming from your blogs.”
Then in Error Reporting, he found the root cause
“Quota is updated. All fixed. :-)” typed SClaus.

Who this merry DevOps elf was, I never shall know.
For before we did our postmortem, away did he go.
Just before vanishing, he took time to write,
“Merry monitoring to all and to all a silent night!”
Happy holidays everyone, and see you in 2017!

December 25, 2016
Highly Anticipated Oculus Touch VR Controllers Now For Sale

Facebook has announced that the highly anticipated Oculus Touch VR Controllers are now available for purchase. Touch allows you to use your hand gestures intuitively within the virtual world, making for a much more immersive experience.

Facebook is also releasing 54 games and experiences today, each designed to make immersion and social interaction more authentic and to give users a reason to order the new Oculus Touch.

They say it’s different by design:

Thanks to Touch’s intuitive and ergonomic design, you forget the controllers and feel like your physical and virtual hands are identical.

Precise tracking and ergonomic handle design work in tandem to bring hand poses and social gestures into VR for a more immersive experience. The diagonal rather than vertical grip lets your thumb and index finger move autonomously, which helps prevent accidental triggers and puts you in control. Thanks to balanced weight distribution, Touch feels natural in your hand so you can relax and enjoy the virtual world.

Hand presence opens up new opportunities to interact with others while experiencing VR—and this is just the first step. We can’t wait to see the next wave of immersive content made possible with Touch.

— The Oculus Team

What people are saying…

“I got mine yesterday, its money well spent, its far better than I imagined it could be,” said David Batty, a Director at Code College, his own startup. “The first contact demo is excellent as an intro to it.”

“You people are doin gods work,” says Daron Thomas. While Nanci Schwartz says, “I’m not sure what planet I’m on anymore. OMG.”

December 7, 2016
New Amazon F1 Instance Reduces Capital-Intensive and Time-Consuming Steps in App Development

Amazon’s is making news about lots of interesting things at its AWS re:Invent 2016 conference currently underway in Las Vegas, and their just announced AWS F1 Instance is no exception.

“Today we are launching a developer preview of the new F1 instance,” said Jeff Barr, Chief Evangelist at Amazon Web Services. “In addition to building applications and services for your own use, you will be able to package them up for sale and reuse in AWS Marketplace. Putting it all together, you will be able to avoid all of the capital-intensive and time-consuming steps that were once a prerequisite to the use of FPGA-powered applications, using a business model that is more akin to that used for every other type of software. We are giving you the ability to design your own logic, simulate and verify it using cloud-based tools, and then get it to market in a matter of days.”

Here are the specs on the FPGA (there are up to eight of these in a single F1 instance):

– Xilinx UltraScale+ VU9P fabricated using a 16 nm process.
– 64 GiB of ECC-protected memory on a 288-bit wide bus (four DDR4 channels).
– Dedicated PCIe x16 interface to the CPU.
– Approximately 2.5 million logic elements.
– Approximately 6,800 Digital Signal Processing (DSP) engines.
– Virtual JTAG interface for debugging.

The F1 instance will significantly speed up applications that are built for a specific purpose. “The general purpose tools can be used to solve many different problems, but may not be the best choice for any particular one,” says Barr. “Purpose-built tools excel at one task, but you may need to do that particular task infrequently.”

Typically says Barr this requires another balancing act: trading off the potential for incredible performance vs. a development life cycle often measured in quarters or years.

“One of the more interesting routes to a custom, hardware-based solution is known as a Field Programmable Gate Array, or FPGA,” said Barr. ”
This highly parallelized model is ideal for building custom accelerators to process compute-intensive problems. Properly programmed, an FPGA has the potential to provide a 30x speedup to many types of genomics, seismic analysis, financial risk analysis, big data search, and encryption algorithms and applications.”

“I hope that this sounds awesome and that you are chomping at the bit to use FPGAs to speed up your own applications,” said Barr. “There are a few interesting challenges along the way. First, FPGAs have traditionally been a component of a larger, purpose-built system. You cannot simply buy one and plug it in to your desktop. Instead, the route to FPGA-powered solutions has included hardware prototyping, construction of a hardware appliance, mass production, and a lengthy sales & deployment cycle. The lead time can limit the applicability of FPGAs, and also means that Moore’s Law has time to make CPU-based solutions more cost-effective.”

Amazon believes that they can do better, and that’s where the F1 instance comes in.

“The bottom line here is that the combination of the F1 instances, the cloud-based development tools, and the ability to sell FPGA-powered applications is unique and powerful,” says Barr. “The power and flexibility of the FPGA model is now accessible all AWS users; I am sure that this will inspire entirely new types of applications and businesses.”

Developers can sign up now for the Amazon EC2 F1 Instances (Preview).

November 30, 2016