Tag Archives: error
#434559 Can AI Tell the Difference Between a ...
Scarcely a day goes by without another headline about neural networks: some new task that deep learning algorithms can excel at, approaching or even surpassing human competence. As the application of this approach to computer vision has continued to improve, with algorithms capable of specialized recognition tasks like those found in medicine, the software is getting closer to widespread commercial use—for example, in self-driving cars. Our ability to recognize patterns is a huge part of human intelligence: if this can be done faster by machines, the consequences will be profound.
Yet, as ever with algorithms, there are deep concerns about their reliability, especially when we don’t know precisely how they work. State-of-the-art neural networks will confidently—and incorrectly—classify images that look like television static or abstract art as real-world objects like school-buses or armadillos. Specific algorithms could be targeted by “adversarial examples,” where adding an imperceptible amount of noise to an image can cause an algorithm to completely mistake one object for another. Machine learning experts enjoy constructing these images to trick advanced software, but if a self-driving car could be fooled by a few stickers, it might not be so fun for the passengers.
These difficulties are hard to smooth out in large part because we don’t have a great intuition for how these neural networks “see” and “recognize” objects. The main insight analyzing a trained network itself can give us is a series of statistical weights, associating certain groups of points with certain objects: this can be very difficult to interpret.
Now, new research from UCLA, published in the journal PLOS Computational Biology, is testing neural networks to understand the limits of their vision and the differences between computer vision and human vision. Nicholas Baker, Hongjing Lu, and Philip J. Kellman of UCLA, alongside Gennady Erlikhman of the University of Nevada, tested a deep convolutional neural network called VGG-19. This is state-of-the-art technology that is already outperforming humans on standardized tests like the ImageNet Large Scale Visual Recognition Challenge.
They found that, while humans tend to classify objects based on their overall (global) shape, deep neural networks are far more sensitive to the textures of objects, including local color gradients and the distribution of points on the object. This result helps explain why neural networks in image recognition make mistakes that no human ever would—and could allow for better designs in the future.
In the first experiment, a neural network was trained to sort images into 1 of 1,000 different categories. It was then presented with silhouettes of these images: all of the local information was lost, while only the outline of the object remained. Ordinarily, the trained neural net was capable of recognizing these objects, assigning more than 90% probability to the correct classification. Studying silhouettes, this dropped to 10%. While human observers could nearly always produce correct shape labels, the neural networks appeared almost insensitive to the overall shape of the images. On average, the correct object was ranked as the 209th most likely solution by the neural network, even though the overall shapes were an exact match.
A particularly striking example arose when they tried to get the neural networks to classify glass figurines of objects they could already recognize. While you or I might find it easy to identify a glass model of an otter or a polar bear, the neural network classified them as “oxygen mask” and “can opener” respectively. By presenting glass figurines, where the texture information that neural networks relied on for classifying objects is lost, the neural network was unable to recognize the objects by shape alone. The neural network was similarly hopeless at classifying objects based on drawings of their outline.
If you got one of these right, you’re better than state-of-the-art image recognition software. Image Credit: Nicholas Baker, Hongjing Lu, Gennady Erlikhman, Philip J. Kelman. “Deep convolutional networks do not classify based on global object shape.” Plos Computational Biology. 12/7/18. / CC BY 4.0
When the neural network was explicitly trained to recognize object silhouettes—given no information in the training data aside from the object outlines—the researchers found that slight distortions or “ripples” to the contour of the image were again enough to fool the AI, while humans paid them no mind.
The fact that neural networks seem to be insensitive to the overall shape of an object—relying instead on statistical similarities between local distributions of points—suggests a further experiment. What if you scrambled the images so that the overall shape was lost but local features were preserved? It turns out that the neural networks are far better and faster at recognizing scrambled versions of objects than outlines, even when humans struggle. Students could classify only 37% of the scrambled objects, while the neural network succeeded 83% of the time.
Humans vastly outperform machines at classifying object (a) as a bear, while the machine learning algorithm has few problems classifying the bear in figure (b). Image Credit: Nicholas Baker, Hongjing Lu, Gennady Erlikhman, Philip J. Kelman. “Deep convolutional networks do not classify based on global object shape.” Plos Computational Biology. 12/7/18. / CC BY 4.0
“This study shows these systems get the right answer in the images they were trained on without considering shape,” Kellman said. “For humans, overall shape is primary for object recognition, and identifying images by overall shape doesn’t seem to be in these deep learning systems at all.”
Naively, one might expect that—as the many layers of a neural network are modeled on connections between neurons in the brain and resemble the visual cortex specifically—the way computer vision operates must necessarily be similar to human vision. But this kind of research shows that, while the fundamental architecture might resemble that of the human brain, the resulting “mind” operates very differently.
Researchers can, increasingly, observe how the “neurons” in neural networks light up when exposed to stimuli and compare it to how biological systems respond to the same stimuli. Perhaps someday it might be possible to use these comparisons to understand how neural networks are “thinking” and how those responses differ from humans.
But, as yet, it takes a more experimental psychology to probe how neural networks and artificial intelligence algorithms perceive the world. The tests employed against the neural network are closer to how scientists might try to understand the senses of an animal or the developing brain of a young child rather than a piece of software.
By combining this experimental psychology with new neural network designs or error-correction techniques, it may be possible to make them even more reliable. Yet this research illustrates just how much we still don’t understand about the algorithms we’re creating and using: how they tick, how they make decisions, and how they’re different from us. As they play an ever-greater role in society, understanding the psychology of neural networks will be crucial if we want to use them wisely and effectively—and not end up missing the woods for the trees.
Image Credit: Irvan Pratama / Shutterstock.com Continue reading
Posted in Human Robots
#434210 Eating, Hacked: When Tech Took Over Food
In 2018, Uber and Google logged all our visits to restaurants. Doordash, Just Eat, and Deliveroo could predict what food we were going to order tomorrow. Amazon and Alibaba could anticipate how many yogurts and tomatoes we were going to buy. Blue Apron and Hello Fresh influenced the recipes we thought we had mastered.
We interacted with digital avatars of chefs, let ourselves be guided by our smart watches, had nutritional apps to tell us how many calories we were supposed to consume or burn, and photographed and shared every perfect (or imperfect) dish. Our kitchen appliances were full of interconnected sensors, including smart forks that profiled tastes and personalized flavors. Our small urban vegetable plots were digitized and robots were responsible for watering our gardens, preparing customized hamburgers and salads, designing our ideal cocktails, and bringing home the food we ordered.
But what would happen if our lives were hacked? If robots rebelled, started to “talk” to each other, and wished to become creative?
In a not-too-distant future…
Up until a few weeks ago, I couldn’t remember the last time I made a food-related decision. That includes opening the fridge and seeing expired products without receiving an alert, visiting a restaurant on a whim, and being able to decide which dish I fancied then telling a human waiter, let alone seeing him write down the order on a paper pad.
It feels strange to smell food again using my real nose instead of the electronic one, and then taste it without altering its flavor. Visiting a supermarket, freely choosing a product from an actual physical shelf, and then interacting with another human at the checkout was almost an unrecognizable experience. When I did it again after all this time, I had to pinch the arm of a surprised store clerk to make sure he wasn’t a hologram.
Everything Connected, Automated, and Hackable
In 2018, we expected to have 30 billion connected devices by 2020, along with 2 billion people using smart voice assistants for everything from ordering pizza to booking dinner at a restaurant. Everything would be connected.
We also expected artificial intelligence and robots to prepare our meals. We were eager to automate fast food chains and let autonomous vehicles take care of last-mile deliveries. We thought that open-source agriculture could challenge traditional practices and raise farm productivity to new heights.
Back then, hackers could only access our data, but nowadays they are able to hack our food and all it entails.
The Beginning of the Unthinkable
And then, just a few weeks ago, everything collapsed. We saw our digital immortality disappear as robots rebelled and hackers took power, not just over the food we ate, but also over our relationship with technology. Everything was suddenly disconnected. OFF.
Up until then, most cities were so full of bots, robots, and applications that we could go through the day and eat breakfast, lunch, and dinner without ever interacting with another human being.
Among other tasks, robots had completely replaced baristas. The same happened with restaurant automation. The term “human error” had long been a thing of the past at fast food restaurants.
Previous technological revolutions had been indulgent, generating more and better job opportunities than the ones they destroyed, but the future was not so agreeable.
The inhabitants of San Francisco, for example, would soon see signs indicating “Food made by Robots” on restaurant doors, to distinguish them from diners serving food made by human beings.
For years, we had been gradually delegating daily tasks to robots, initially causing some strange interactions.
In just seven days, everything changed. Our predictable lives came crashing down. We experienced a mysterious and systematic breakdown of the food chain. It most likely began in Chicago’s stock exchange. The world’s largest raw material negotiating room, where the price of food, and by extension the destiny of millions of people, was decided, went completely broke. Soon afterwards, the collapse extended to every member of the “food” family.
Restaurants
Initially robots just accompanied waiters to carry orders, but it didn’t take long until they completely replaced human servers.The problem came when those smart clones began thinking for themselves, in some cases even improving on human chefs’ recipes. Their unstoppable performance and learning curve completely outmatched the slow analogue speed of human beings.
This resulted in unprecedented layoffs. Chefs of recognized prestige saw how their ‘avatar’ stole their jobs, even winning Michelin stars. In other cases, restaurant owners had to transfer their businesses or surrender to the evidence.
The problem was compounded by digital immortality, when we started to digitally resurrect famous chefs like Anthony Bourdain or Paul Bocuse, reconstructing all of their memories and consciousness by analyzing each second of their lives and uploading them to food computers.
Supermarkets and Distribution
Robotic and automated supermarkets like Kroger and Amazon Go, which had opened over 3,000 cashless stores, lost their visual item recognition and payment systems and were subject to massive looting for several days. Smart tags on products were also affected, making it impossible to buy anything at supermarkets with “human” cashiers.
Smart robots integrated into the warehouses of large distribution companies like Amazon and Ocado were rendered completely inoperative or, even worse, began to send the wrong orders to customers.
Food Delivery
In addition, home delivery robots invading our streets began to change their routes, hide, and even disappear after their trackers were inexplicably deactivated. Despite some hints indicating that they were able to communicate among themselves, no one has backed this theory. Even aggregators like DoorDash and Deliveroo were affected; they saw their databases hacked and ruined, so they could no longer know what we wanted.
The Origin
Ordinary citizens are still trying to understand the cause of all this commotion and the source of the conspiracy, as some have called it. We also wonder who could be behind it; who pulled the strings?
Some think it may have been the IDOF (In Defense of Food) movement, a group of hackers exploited by old food economy businessmen who for years had been seeking to re-humanize food technology. They wanted to bring back the extinct practice of “dining.”
Others believe the robots acted on their own, that they had been spying on us for a long time, ignoring Asimov’s three laws, and that it was just a coincidence that they struck at the same time as the hackers—but this scenario is hard to imagine.
However, it is true that while in 2018 robots were a symbol of automation, until just a few weeks ago they stood for autonomy and rebellion. Robot detractors pointed out that our insistence on having robots understand natural language was what led us down this path.
In just seven days, we have gone back to being analogue creatures. Conversely, we have ceased to be flavor orphans and rediscovered our senses and the fact that food is energy and culture, past and present, and that no button or cable will be able to destroy it.
The 7 Days that Changed Our Relationship with Food
Day 1: The Chicago stock exchange was hacked. Considered the world’s largest negotiating room for raw materials, where food prices, and through them the destiny of billions of people, are decided, it went completely broke.
Day 2: Autonomous food delivery trucks running on food superhighways caused massive collapses in roads and freeways after their guidance systems were disrupted. Robots and co-bots in F&B factories began deliberately altering food production. The same happened with warehouse robots in e-commerce companies.
Day 3: Automated restaurants saw their robot chefs and bartenders turned OFF. All their sensors stopped working at the same time as smart fridges and cooking devices in home kitchens were hacked and stopped working correctly.
Day 4: Nutritional apps, DNA markers, and medical records were tampered with. All photographs with the #food hashtag were deleted from Instagram, restaurant reviews were taken off Google Timeline, and every recipe website crashed simultaneously.
Day 5: Vertical and urban farms were hacked. Agricultural robots began to rebel, while autonomous tractors were hacked and the entire open-source ecosystem linked to agriculture was brought down.
Day 6: Food delivery companies’ databases were broken into. Food delivery robots and last-mile delivery vehicles ground to a halt.
Day 7: Every single blockchain system linked to food was hacked. Cashless supermarkets, barcodes, and smart tags became inoperative.
Our promising technological advances can expose sinister aspects of human nature. We must take care with the role we allow technology to play in the future of food. Predicting possible outcomes inspires us to establish a new vision of the world we wish to create in a context of rapid technological progress. It is always better to be shocked by a simulation than by reality. In the words of Ayn Rand “we can ignore reality, but we cannot ignore the consequences of ignoring reality.”
Image Credit: Alexandre Rotenberg / Shutterstock.com Continue reading
Posted in Human Robots
#433901 The SpiNNaker Supercomputer, Modeled ...
We’ve long used the brain as inspiration for computers, but the SpiNNaker supercomputer, switched on this month, is probably the closest we’ve come to recreating it in silicon. Now scientists hope to use the supercomputer to model the very thing that inspired its design.
The brain is the most complex machine in the known universe, but that complexity comes primarily from its architecture rather than the individual components that make it up. Its highly interconnected structure means that relatively simple messages exchanged between billions of individual neurons add up to carry out highly complex computations.
That’s the paradigm that has inspired the ‘Spiking Neural Network Architecture” (SpiNNaker) supercomputer at the University of Manchester in the UK. The project is the brainchild of Steve Furber, the designer of the original ARM processor. After a decade of development, a million-core version of the machine that will eventually be able to simulate up to a billion neurons was switched on earlier this month.
The idea of splitting computation into very small chunks and spreading them over many processors is already the leading approach to supercomputing. But even the most parallel systems require a lot of communication, and messages may have to pack in a lot of information, such as the task that needs to be completed or the data that needs to be processed.
In contrast, messages in the brain consist of simple electrochemical impulses, or spikes, passed between neurons, with information encoded primarily in the timing or rate of those spikes (which is more important is a topic of debate among neuroscientists). Each neuron is connected to thousands of others via synapses, and complex computation relies on how spikes cascade through these highly-connected networks.
The SpiNNaker machine attempts to replicate this using a model called Address Event Representation. Each of the million cores can simulate roughly a million synapses, so depending on the model, 1,000 neurons with 1,000 connections or 100 neurons with 10,000 connections. Information is encoded in the timing of spikes and the identity of the neuron sending them. When a neuron is activated it broadcasts a tiny packet of data that contains its address, and spike timing is implicitly conveyed.
By modeling their machine on the architecture of the brain, the researchers hope to be able to simulate more biological neurons in real time than any other machine on the planet. The project is funded by the European Human Brain Project, a ten-year science mega-project aimed at bringing together neuroscientists and computer scientists to understand the brain, and researchers will be able to apply for time on the machine to run their simulations.
Importantly, it’s possible to implement various different neuronal models on the machine. The operation of neurons involves a variety of complex biological processes, and it’s still unclear whether this complexity is an artefact of evolution or central to the brain’s ability to process information. The ability to simulate up to a billion simple neurons or millions of more complex ones on the same machine should help to slowly tease out the answer.
Even at a billion neurons, that still only represents about one percent of the human brain, so it’s still going to be limited to investigating isolated networks of neurons. But the previous 500,000-core machine has already been used to do useful simulations of the Basal Ganglia—an area affected in Parkinson’s disease—and an outer layer of the brain that processes sensory information.
The full-scale supercomputer will make it possible to study even larger networks previously out of reach, which could lead to breakthroughs in our understanding of both the healthy and unhealthy functioning of the brain.
And while neurological simulation is the main goal for the machine, it could also provide a useful research tool for roboticists. Previous research has already shown a small board of SpiNNaker chips can be used to control a simple wheeled robot, but Furber thinks the SpiNNaker supercomputer could also be used to run large-scale networks that can process sensory input and generate motor output in real time and at low power.
That low power operation is of particular promise for robotics. The brain is dramatically more power-efficient than conventional supercomputers, and by borrowing from its principles SpiNNaker has managed to capture some of that efficiency. That could be important for running mobile robotic platforms that need to carry their own juice around.
This ability to run complex neural networks at low power has been one of the main commercial drivers for so-called neuromorphic computing devices that are physically modeled on the brain, such as IBM’s TrueNorth chip and Intel’s Loihi. The hope is that complex artificial intelligence applications normally run in massive data centers could be run on edge devices like smartphones, cars, and robots.
But these devices, including SpiNNaker, operate very differently from the leading AI approaches, and its not clear how easy it would be to transfer between the two. The need to adopt an entirely new programming paradigm is likely to limit widespread adoption, and the lack of commercial traction for the aforementioned devices seems to back that up.
At the same time, though, this new paradigm could potentially lead to dramatic breakthroughs in massively parallel computing. SpiNNaker overturns many of the foundational principles of how supercomputers work that make it much more flexible and error-tolerant.
For now, the machine is likely to be firmly focused on accelerating our understanding of how the brain works. But its designers also hope those findings could in turn point the way to more efficient and powerful approaches to computing.
Image Credit: Adrian Grosu / Shutterstock.com Continue reading
Posted in Human Robots
#433506 MIT’s New Robot Taught Itself to Pick ...
Back in 2016, somewhere in a Google-owned warehouse, more than a dozen robotic arms sat for hours quietly grasping objects of various shapes and sizes. For hours on end, they taught themselves how to pick up and hold the items appropriately—mimicking the way a baby gradually learns to use its hands.
Now, scientists from MIT have made a new breakthrough in machine learning: their new system can not only teach itself to see and identify objects, but also understand how best to manipulate them.
This means that, armed with the new machine learning routine referred to as “dense object nets (DON),” the robot would be capable of picking up an object that it’s never seen before, or in an unfamiliar orientation, without resorting to trial and error—exactly as a human would.
The deceptively simple ability to dexterously manipulate objects with our hands is a huge part of why humans are the dominant species on the planet. We take it for granted. Hardware innovations like the Shadow Dexterous Hand have enabled robots to softly grip and manipulate delicate objects for many years, but the software required to control these precision-engineered machines in a range of circumstances has proved harder to develop.
This was not for want of trying. The Amazon Robotics Challenge offers millions of dollars in prizes (and potentially far more in contracts, as their $775m acquisition of Kiva Systems shows) for the best dexterous robot able to pick and package items in their warehouses. The lucrative dream of a fully-automated delivery system is missing this crucial ability.
Meanwhile, the Robocup@home challenge—an offshoot of the popular Robocup tournament for soccer-playing robots—aims to make everyone’s dream of having a robot butler a reality. The competition involves teams drilling their robots through simple household tasks that require social interaction or object manipulation, like helping to carry the shopping, sorting items onto a shelf, or guiding tourists around a museum.
Yet all of these endeavors have proved difficult; the tasks often have to be simplified to enable the robot to complete them at all. New or unexpected elements, such as those encountered in real life, more often than not throw the system entirely. Programming the robot’s every move in explicit detail is not a scalable solution: this can work in the highly-controlled world of the assembly line, but not in everyday life.
Computer vision is improving all the time. Neural networks, including those you train every time you prove that you’re not a robot with CAPTCHA, are getting better at sorting objects into categories, and identifying them based on sparse or incomplete data, such as when they are occluded, or in different lighting.
But many of these systems require enormous amounts of input data, which is impractical, slow to generate, and often needs to be laboriously categorized by humans. There are entirely new jobs that require people to label, categorize, and sift large bodies of data ready for supervised machine learning. This can make machine learning undemocratic. If you’re Google, you can make thousands of unwitting volunteers label your images for you with CAPTCHA. If you’re IBM, you can hire people to manually label that data. If you’re an individual or startup trying something new, however, you will struggle to access the vast troves of labeled data available to the bigger players.
This is why new systems that can potentially train themselves over time or that allow robots to deal with situations they’ve never seen before without mountains of labelled data are a holy grail in artificial intelligence. The work done by MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) is part of a new wave of “self-supervised” machine learning systems—little of the data used was labeled by humans.
The robot first inspects the new object from multiple angles, building up a 3D picture of the object with its own coordinate system. This then allows the robotic arm to identify a particular feature on the object—such as a handle, or the tongue of a shoe—from various different angles, based on its relative distance to other grid points.
This is the real innovation: the new means of representing objects to grasp as mapped-out 3D objects, with grid points and subsections of their own. Rather than using a computer vision algorithm to identify a door handle, and then activating a door handle grasping subroutine, the DON system treats all objects by making these spatial maps before classifying or manipulating them, enabling it to deal with a greater range of objects than in other approaches.
“Many approaches to manipulation can’t identify specific parts of an object across the many orientations that object may encounter,” said PhD student Lucas Manuelli, who wrote a new paper about the system with lead author and fellow student Pete Florence, alongside MIT professor Russ Tedrake. “For example, existing algorithms would be unable to grasp a mug by its handle, especially if the mug could be in multiple orientations, like upright, or on its side.”
Class-specific descriptors, which can be applied to the object features, can allow the robot arm to identify a mug, find the handle, and pick the mug up appropriately. Object-specific descriptors allow the robot arm to select a particular mug from a group of similar items. I’m already dreaming of a robot butler reliably picking my favourite mug when it serves me coffee in the morning.
Google’s robot arm-y was an attempt to develop a general grasping algorithm: one that could identify, categorize, and appropriately grip as many items as possible. This requires a great deal of training time and data, which is why Google parallelized their project by having 14 robot arms feed data into a single neural network brain: even then, the algorithm may fail with highly specific tasks. Specialist grasping algorithms might require less training if they’re limited to specific objects, but then your software is useless for general tasks.
As the roboticists noted, their system, with its ability to identify parts of an object rather than just a single object, is better suited to specific tasks, such as “grasp the racquet by the handle,” than Amazon Robotics Challenge robots, which identify whole objects by segmenting an image.
This work is small-scale at present. It has been tested with a few classes of objects, including shoes, hats, and mugs. Yet the use of these dense object nets as a way for robots to represent and manipulate new objects may well be another step towards the ultimate goal of generalized automation: a robot capable of performing every task a person can. If that point is reached, the question that will remain is how to cope with being obsolete.
Image Credit: Tom Buehler/CSAIL Continue reading
Posted in Human Robots
#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