Tag Archives: automatic

#431872 AI Uses Titan Supercomputer to Create ...

You don’t have to dig too deeply into the archive of dystopian science fiction to uncover the horror that intelligent machines might unleash. The Matrix and The Terminator are probably the most well-known examples of self-replicating, intelligent machines attempting to enslave or destroy humanity in the process of building a brave new digital world.
The prospect of artificially intelligent machines creating other artificially intelligent machines took a big step forward in 2017. However, we’re far from the runaway technological singularity futurists are predicting by mid-century or earlier, let alone murderous cyborgs or AI avatar assassins.
The first big boost this year came from Google. The tech giant announced it was developing automated machine learning (AutoML), writing algorithms that can do some of the heavy lifting by identifying the right neural networks for a specific job. Now researchers at the Department of Energy’s Oak Ridge National Laboratory (ORNL), using the most powerful supercomputer in the US, have developed an AI system that can generate neural networks as good if not better than any developed by a human in less than a day.
It can take months for the brainiest, best-paid data scientists to develop deep learning software, which sends data through a complex web of mathematical algorithms. The system is modeled after the human brain and known as an artificial neural network. Even Google’s AutoML took weeks to design a superior image recognition system, one of the more standard operations for AI systems today.
Computing Power
Of course, Google Brain project engineers only had access to 800 graphic processing units (GPUs), a type of computer hardware that works especially well for deep learning. Nvidia, which pioneered the development of GPUs, is considered the gold standard in today’s AI hardware architecture. Titan, the supercomputer at ORNL, boasts more than 18,000 GPUs.
The ORNL research team’s algorithm, called MENNDL for Multinode Evolutionary Neural Networks for Deep Learning, isn’t designed to create AI systems that cull cute cat photos from the internet. Instead, MENNDL is a tool for testing and training thousands of potential neural networks to work on unique science problems.
That requires a different approach from the Google and Facebook AI platforms of the world, notes Steven Young, a postdoctoral research associate at ORNL who is on the team that designed MENNDL.
“We’ve discovered that those [neural networks] are very often not the optimal network for a lot of our problems, because our data, while it can be thought of as images, is different,” he explains to Singularity Hub. “These images, and the problems, have very different characteristics from object detection.”
AI for Science
One application of the technology involved a particle physics experiment at the Fermi National Accelerator Laboratory. Fermilab researchers are interested in understanding neutrinos, high-energy subatomic particles that rarely interact with normal matter but could be a key to understanding the early formation of the universe. One Fermilab experiment involves taking a sort of “snapshot” of neutrino interactions.
The team wanted the help of an AI system that could analyze and classify Fermilab’s detector data. MENNDL evaluated 500,000 neural networks in 24 hours. Its final solution proved superior to custom models developed by human scientists.
In another case involving a collaboration with St. Jude Children’s Research Hospital in Memphis, MENNDL improved the error rate of a human-designed algorithm for identifying mitochondria inside 3D electron microscopy images of brain tissue by 30 percent.
“We are able to do better than humans in a fraction of the time at designing networks for these sort of very different datasets that we’re interested in,” Young says.
What makes MENNDL particularly adept is its ability to define the best or most optimal hyperparameters—the key variables—to tackle a particular dataset.
“You don’t always need a big, huge deep network. Sometimes you just need a small network with the right hyperparameters,” Young says.
A Virtual Data Scientist
That’s not dissimilar to the approach of a company called H20.ai, a startup out of Silicon Valley that uses open source machine learning platforms to “democratize” AI. It applies machine learning to create business solutions for Fortune 500 companies, including some of the world’s biggest banks and healthcare companies.
“Our software is more [about] pattern detection, let’s say anti-money laundering or fraud detection or which customer is most likely to churn,” Dr. Arno Candel, chief technology officer at H2O.ai, tells Singularity Hub. “And that kind of insight-generating software is what we call AI here.”
The company’s latest product, Driverless AI, promises to deliver the data scientist equivalent of a chessmaster to its customers (the company claims several such grandmasters in its employ and advisory board). In other words, the system can analyze a raw dataset and, like MENNDL, automatically identify what features should be included in the computer model to make the most of the data based on the best “chess moves” of its grandmasters.
“So we’re using those algorithms, but we’re giving them the human insights from those data scientists, and we automate their thinking,” he explains. “So we created a virtual data scientist that is relentless at trying these ideas.”
Inside the Black Box
Not unlike how the human brain reaches a conclusion, it’s not always possible to understand how a machine, despite being designed by humans, reaches its own solutions. The lack of transparency is often referred to as the AI “black box.” Experts like Young say we can learn something about the evolutionary process of machine learning by generating millions of neural networks and seeing what works well and what doesn’t.
“You’re never going to be able to completely explain what happened, but maybe we can better explain it than we currently can today,” Young says.
Transparency is built into the “thought process” of each particular model generated by Driverless AI, according to Candel.
The computer even explains itself to the user in plain English at each decision point. There is also real-time feedback that allows users to prioritize features, or parameters, to see how the changes improve the accuracy of the model. For example, the system may include data from people in the same zip code as it creates a model to describe customer turnover.
“That’s one of the advantages of our automatic feature engineering: it’s basically mimicking human thinking,” Candel says. “It’s not just neural nets that magically come up with some kind of number, but we’re trying to make it statistically significant.”
Moving Forward
Much digital ink has been spilled over the dearth of skilled data scientists, so automating certain design aspects for developing artificial neural networks makes sense. Experts agree that automation alone won’t solve that particular problem. However, it will free computer scientists to tackle more difficult issues, such as parsing the inherent biases that exist within the data used by machine learning today.
“I think the world has an opportunity to focus more on the meaning of things and not on the laborious tasks of just fitting a model and finding the best features to make that model,” Candel notes. “By automating, we are pushing the burden back for the data scientists to actually do something more meaningful, which is think about the problem and see how you can address it differently to make an even bigger impact.”
The team at ORNL expects it can also make bigger impacts beginning next year when the lab’s next supercomputer, Summit, comes online. While Summit will boast only 4,600 nodes, it will sport the latest and greatest GPU technology from Nvidia and CPUs from IBM. That means it will deliver more than five times the computational performance of Titan, the world’s fifth-most powerful supercomputer today.
“We’ll be able to look at much larger problems on Summit than we were able to with Titan and hopefully get to a solution much faster,” Young says.
It’s all in a day’s work.
Image Credit: Gennady Danilkin / Shutterstock.com Continue reading

Posted in Human Robots

#431599 8 Ways AI Will Transform Our Cities by ...

How will AI shape the average North American city by 2030? A panel of experts assembled as part of a century-long study into the impact of AI thinks its effects will be profound.
The One Hundred Year Study on Artificial Intelligence is the brainchild of Eric Horvitz, technical fellow and a managing director at Microsoft Research.
Every five years a panel of experts will assess the current state of AI and its future directions. The first panel, comprised of experts in AI, law, political science, policy, and economics, was launched last fall and decided to frame their report around the impact AI will have on the average American city. Here’s how they think it will affect eight key domains of city life in the next fifteen years.
1. Transportation
The speed of the transition to AI-guided transport may catch the public by surprise. Self-driving vehicles will be widely adopted by 2020, and it won’t just be cars — driverless delivery trucks, autonomous delivery drones, and personal robots will also be commonplace.
Uber-style “cars as a service” are likely to replace car ownership, which may displace public transport or see it transition towards similar on-demand approaches. Commutes will become a time to relax or work productively, encouraging people to live further from home, which could combine with reduced need for parking to drastically change the face of modern cities.
Mountains of data from increasing numbers of sensors will allow administrators to model individuals’ movements, preferences, and goals, which could have major impact on the design city infrastructure.
Humans won’t be out of the loop, though. Algorithms that allow machines to learn from human input and coordinate with them will be crucial to ensuring autonomous transport operates smoothly. Getting this right will be key as this will be the public’s first experience with physically embodied AI systems and will strongly influence public perception.
2. Home and Service Robots
Robots that do things like deliver packages and clean offices will become much more common in the next 15 years. Mobile chipmakers are already squeezing the power of last century’s supercomputers into systems-on-a-chip, drastically boosting robots’ on-board computing capacity.
Cloud-connected robots will be able to share data to accelerate learning. Low-cost 3D sensors like Microsoft’s Kinect will speed the development of perceptual technology, while advances in speech comprehension will enhance robots’ interactions with humans. Robot arms in research labs today are likely to evolve into consumer devices around 2025.
But the cost and complexity of reliable hardware and the difficulty of implementing perceptual algorithms in the real world mean general-purpose robots are still some way off. Robots are likely to remain constrained to narrow commercial applications for the foreseeable future.
3. Healthcare
AI’s impact on healthcare in the next 15 years will depend more on regulation than technology. The most transformative possibilities of AI in healthcare require access to data, but the FDA has failed to find solutions to the difficult problem of balancing privacy and access to data. Implementation of electronic health records has also been poor.
If these hurdles can be cleared, AI could automate the legwork of diagnostics by mining patient records and the scientific literature. This kind of digital assistant could allow doctors to focus on the human dimensions of care while using their intuition and experience to guide the process.
At the population level, data from patient records, wearables, mobile apps, and personal genome sequencing will make personalized medicine a reality. While fully automated radiology is unlikely, access to huge datasets of medical imaging will enable training of machine learning algorithms that can “triage” or check scans, reducing the workload of doctors.
Intelligent walkers, wheelchairs, and exoskeletons will help keep the elderly active while smart home technology will be able to support and monitor them to keep them independent. Robots may begin to enter hospitals carrying out simple tasks like delivering goods to the right room or doing sutures once the needle is correctly placed, but these tasks will only be semi-automated and will require collaboration between humans and robots.
4. Education
The line between the classroom and individual learning will be blurred by 2030. Massive open online courses (MOOCs) will interact with intelligent tutors and other AI technologies to allow personalized education at scale. Computer-based learning won’t replace the classroom, but online tools will help students learn at their own pace using techniques that work for them.
AI-enabled education systems will learn individuals’ preferences, but by aggregating this data they’ll also accelerate education research and the development of new tools. Online teaching will increasingly widen educational access, making learning lifelong, enabling people to retrain, and increasing access to top-quality education in developing countries.
Sophisticated virtual reality will allow students to immerse themselves in historical and fictional worlds or explore environments and scientific objects difficult to engage with in the real world. Digital reading devices will become much smarter too, linking to supplementary information and translating between languages.
5. Low-Resource Communities
In contrast to the dystopian visions of sci-fi, by 2030 AI will help improve life for the poorest members of society. Predictive analytics will let government agencies better allocate limited resources by helping them forecast environmental hazards or building code violations. AI planning could help distribute excess food from restaurants to food banks and shelters before it spoils.
Investment in these areas is under-funded though, so how quickly these capabilities will appear is uncertain. There are fears valueless machine learning could inadvertently discriminate by correlating things with race or gender, or surrogate factors like zip codes. But AI programs are easier to hold accountable than humans, so they’re more likely to help weed out discrimination.
6. Public Safety and Security
By 2030 cities are likely to rely heavily on AI technologies to detect and predict crime. Automatic processing of CCTV and drone footage will make it possible to rapidly spot anomalous behavior. This will not only allow law enforcement to react quickly but also forecast when and where crimes will be committed. Fears that bias and error could lead to people being unduly targeted are justified, but well-thought-out systems could actually counteract human bias and highlight police malpractice.
Techniques like speech and gait analysis could help interrogators and security guards detect suspicious behavior. Contrary to concerns about overly pervasive law enforcement, AI is likely to make policing more targeted and therefore less overbearing.
7. Employment and Workplace
The effects of AI will be felt most profoundly in the workplace. By 2030 AI will be encroaching on skilled professionals like lawyers, financial advisers, and radiologists. As it becomes capable of taking on more roles, organizations will be able to scale rapidly with relatively small workforces.
AI is more likely to replace tasks rather than jobs in the near term, and it will also create new jobs and markets, even if it’s hard to imagine what those will be right now. While it may reduce incomes and job prospects, increasing automation will also lower the cost of goods and services, effectively making everyone richer.
These structural shifts in the economy will require political rather than purely economic responses to ensure these riches are shared. In the short run, this may include resources being pumped into education and re-training, but longer term may require a far more comprehensive social safety net or radical approaches like a guaranteed basic income.
8. Entertainment
Entertainment in 2030 will be interactive, personalized, and immeasurably more engaging than today. Breakthroughs in sensors and hardware will see virtual reality, haptics and companion robots increasingly enter the home. Users will be able to interact with entertainment systems conversationally, and they will show emotion, empathy, and the ability to adapt to environmental cues like the time of day.
Social networks already allow personalized entertainment channels, but the reams of data being collected on usage patterns and preferences will allow media providers to personalize entertainment to unprecedented levels. There are concerns this could endow media conglomerates with unprecedented control over people’s online experiences and the ideas to which they are exposed.
But advances in AI will also make creating your own entertainment far easier and more engaging, whether by helping to compose music or choreograph dances using an avatar. Democratizing the production of high-quality entertainment makes it nearly impossible to predict how highly fluid human tastes for entertainment will develop.
Image Credit: Asgord / Shutterstock.com Continue reading

Posted in Human Robots

#431175 Servosila introduces Mobile Robots ...

Servosila introduces a new member of the family of Servosila “Engineer” robots, a UGV called “Radio Engineer”. This new variant of the well-known backpack-transportable robot features a Software Defined Radio (SDR) payload module integrated into the robotic vehicle.

“Several of our key customers had asked us to enable an Electronic Warfare (EW) or Cognitive Radio applications in our robots”, – says a spokesman for the company, “By integrating a Software Defined Radio (SDR) module into our robotic platforms we cater to both requirements. Radio spectrum analysis, radio signal detection, jamming, and radio relay are important features for EOD robots such as ours. Servosila continues to serve the customers by pushing the boundaries of what their Servosila robots can do. Our partners in the research world and academia shall also greatly benefit from the new functionality that gives them more means of achieving their research goals.”
Photo Credit: Servosila – www.servosila.com
Coupling a programmable mobile robot with a software-defined radio creates a powerful platform for developing innovative applications that mix mobility and artificial intelligence with modern radio technologies. The new robotic radio applications include localized frequency hopping pattern analysis, OFDM waveform recognition, outdoor signal triangulation, cognitive mesh networking, automatic area search for radio emitters, passive or active mobile robotic radars, mobile base stations, mobile radio scanners, and many others.

A rotating head of the robot with mounts for external antennae acts as a pan-and-tilt device thus enabling various scanning and tracking applications. The neck of the robotic head is equipped with a pair of highly accurate Servosila-made servos with a pointing precision of 3.0 angular minutes. This means that the robot can point its antennae with an unprecedented accuracy.

Researchers and academia can benefit from the platform’s support for GnuRadio, an open source software framework for developing SDR applications. An on-board Intel i7 computer capable of executing OpenCL code, is internally connected to the SDR payload module. This makes it possible to execute most existing GnuRadio applications directly on the robot’s on-board computer. Other sensors of the robot such as a GPS sensor, an IMU or a thermal vision camera contribute into sensor fusion algorithms.

Since Servosila “Engineer” mobile robots are primarily designed for outdoor use, the SDR module is fully enclosed into a hardened body of the robot which provides protection in case of dust, rain, snow or impacts with obstacles while the robot is on the move. The robot and its SDR payload module are both powered by an on-board battery thus making the entire robotic radio platform independent of external power supplies.

Servosila plans to start shipping the SDR-equipped robots to international customers in October, 2017.

Web: https://www.servosila.com
YouTube: https://www.youtube.com/user/servosila/videos

About the Company
Servosila is a robotics technology company that designs, produces and markets a range of mobile robots, robotic arms, servo drives, harmonic reduction gears, robotic control systems as well as software packages that make the robots intelligent. Servosila provides consulting, training and operations support services to various customers around the world. The company markets its products and services directly or through a network of partners who provide tailored and localized services that meet specific procurement, support or operational needs.
Press Release above is by: Servosila
The post Servosila introduces Mobile Robots equipped with Software Defined Radio (SDR) payloads appeared first on Roboticmagazine. Continue reading

Posted in Human Robots

#430855 Why Education Is the Hardest Sector of ...

We’ve all heard the warning cries: automation will disrupt entire industries and put millions of people out of jobs. In fact, up to 45 percent of existing jobs can be automated using current technology.
However, this may not necessarily apply to the education sector. After a detailed analysis of more than 2,000-plus work activities for more than 800 occupations, a report by McKinsey & Co states that of all the sectors examined, “…the technical feasibility of automation is lowest in education.”
There is no doubt that technological trends will have a powerful impact on global education, both by improving the overall learning experience and by increasing global access to education. Massive open online courses (MOOCs), chatbot tutors, and AI-powered lesson plans are just a few examples of the digital transformation in global education. But will robots and artificial intelligence ever fully replace teachers?
The Most Difficult Sector to Automate
While various tasks revolving around education—like administrative tasks or facilities maintenance—are open to automation, teaching itself is not.
Effective education involves more than just transfer of information from a teacher to a student. Good teaching requires complex social interactions and adaptation to the individual student’s learning needs. An effective teacher is not just responsive to each student’s strengths and weaknesses, but is also empathetic towards the student’s state of mind. It’s about maximizing human potential.
Furthermore, students don’t just rely on effective teachers to teach them the course material, but also as a source of life guidance and career mentorship. Deep and meaningful human interaction is crucial and is something that is very difficult, if not impossible, to automate.
Automating teaching is an example of a task that would require artificial general intelligence (as opposed to narrow or specific intelligence). In other words, this is the kind of task that would require an AI that understands natural human language, can be empathetic towards emotions, plan, strategize and make impactful decisions under unpredictable circumstances.
This would be the kind of machine that can do anything a human can do, and it doesn’t exist—at least, not yet.
We’re Getting There
Let’s not forget how quickly AI is evolving. Just because it’s difficult to fully automate teaching, it doesn’t mean the world’s leading AI experts aren’t trying.
Meet Jill Watson, the teaching assistant from Georgia Institute of Technology. Watson isn’t your average TA. She’s an IBM-powered artificial intelligence that is being implemented in universities around the world. Watson is able to answer students’ questions with 97 percent certainty.
Technologies like this also have applications in grading and providing feedback. Some AI algorithms are being trained and refined to perform automatic essay scoring. One project has achieved a 0.945 correlation with human graders.
All of this will have a remarkable impact on online education as we know it and dramatically increase online student retention rates.

Any student with a smartphone can access a wealth of information and free courses from universities around the world. MOOCs have allowed valuable courses to become available to millions of students. But at the moment, not all participants can receive customized feedback for their work. Currently, this is limited by manpower, but in the future that may not be the case.
What chatbots like Jill Watson allow is the opportunity for hundreds of thousands, if not millions, of students to have their work reviewed and all their questions answered at a minimal cost.
AI algorithms also have a significant role to play in personalization of education. Every student is unique and has a different set of strengths and weaknesses. Data analysis can be used to improve individual student results, assess each student’s strengths and weaknesses, and create mass-customized programs. Algorithms can analyze student data and consequently make flexible programs that adapt to the learner based on real-time feedback. According to the McKinsey Global Institute, all of this data in education could unlock between $900 billion and $1.2 trillion in global economic value.
Beyond Automated Teaching
It’s important to recognize that technological automation alone won’t fix the many issues in our global education system today. Dominated by outdated curricula, standardized tests, and an emphasis on short-term knowledge, many experts are calling for a transformation of how we teach.
It is not enough to simply automate the process. We can have a completely digital learning experience that continues to focus on outdated skills and fails to prepare students for the future. In other words, we must not only be innovative with our automation capabilities, but also with educational content, strategy, and policies.
Are we equipping students with the most important survival skills? Are we inspiring young minds to create a better future? Are we meeting the unique learning needs of each and every student? There’s no point automating and digitizing a system that is already flawed. We need to ensure the system that is being digitized is itself being transformed for the better.
Stock Media provided by davincidig / Pond5 Continue reading

Posted in Human Robots