Tag Archives: object

#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 | Tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , | Leave a comment

#431689 Robotic Materials Will Distribute ...

The classical view of a robot as a mechanical body with a central “brain” that controls its behavior could soon be on its way out. The authors of a recent article in Science Robotics argue that future robots will have intelligence distributed throughout their bodies.
The concept, and the emerging discipline behind it, are variously referred to as “material robotics” or “robotic materials” and are essentially a synthesis of ideas from robotics and materials science. Proponents say advances in both fields are making it possible to create composite materials capable of combining sensing, actuation, computation, and communication and operating independently of a central processing unit.
Much of the inspiration for the field comes from nature, with practitioners pointing to the adaptive camouflage of the cuttlefish’s skin, the ability of bird wings to morph in response to different maneuvers, or the banyan tree’s ability to grow roots above ground to support new branches.
Adaptive camouflage and morphing wings have clear applications in the defense and aerospace sector, but the authors say similar principles could be used to create everything from smart tires able to calculate the traction needed for specific surfaces to grippers that can tailor their force to the kind of object they are grasping.
“Material robotics represents an acknowledgment that materials can absorb some of the challenges of acting and reacting to an uncertain world,” the authors write. “Embedding distributed sensors and actuators directly into the material of the robot’s body engages computational capabilities and offloads the rigid information and computational requirements from the central processing system.”
The idea of making materials more adaptive is not new, and there are already a host of “smart materials” that can respond to stimuli like heat, mechanical stress, or magnetic fields by doing things like producing a voltage or changing shape. These properties can be carefully tuned to create materials capable of a wide variety of functions such as movement, self-repair, or sensing.
The authors say synthesizing these kinds of smart materials, alongside other advanced materials like biocompatible conductors or biodegradable elastomers, is foundational to material robotics. But the approach also involves integration of many different capabilities in the same material, careful mechanical design to make the most of mechanical capabilities, and closing the loop between sensing and control within the materials themselves.
While there are stand-alone applications for such materials in the near term, like smart fabrics or robotic grippers, the long-term promise of the field is to distribute decision-making in future advanced robots. As they are imbued with ever more senses and capabilities, these machines will be required to shuttle huge amounts of control and feedback data to and fro, placing a strain on both their communication and computation abilities.
Materials that can process sensor data at the source and either autonomously react to it or filter the most relevant information to be passed on to the central processing unit could significantly ease this bottleneck. In a press release related to an earlier study, Nikolaus Correll, an assistant professor of computer science at the University of Colorado Boulder who is also an author of the current paper, pointed out this is a tactic used by the human body.
“The human sensory system automatically filters out things like the feeling of clothing rubbing on the skin,” he said. “An artificial skin with possibly thousands of sensors could do the same thing, and only report to a central ‘brain’ if it touches something new.”
There are still considerable challenges to realizing this vision, though, the authors say, noting that so far the young field has only produced proof of concepts. The biggest challenge remains manufacturing robotic materials in a way that combines all these capabilities in a small enough package at an affordable cost.
Luckily, the authors note, the field can draw on convergent advances in both materials science, such as the development of new bulk materials with inherent multifunctionality, and robotics, such as the ever tighter integration of components.
And they predict that doing away with the prevailing dichotomy of “brain versus body” could lay the foundations for the emergence of “robots with brains in their bodies—the foundation of inexpensive and ubiquitous robots that will step into the real world.”
Image Credit: Anatomy Insider / Shutterstock.com Continue reading

Posted in Human Robots | Tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , | Leave a comment

#431653 9 Robot Animals Built From Nature’s ...

Millions of years of evolution have allowed animals to develop some elegant and highly efficient solutions to problems like locomotion, flight, and dexterity. As Boston Dynamics unveils its latest mechanical animals, here’s a rundown of nine recent robots that borrow from nature and why.
SpotMini – Boston Dynamics

Starting with BigDog in 2005, the US company has built a whole stable of four-legged robots in recent years. Their first product was designed to be a robotic packhorse for soldiers that borrowed the quadrupedal locomotion of animals to travel over terrain too rough for conventional vehicles.
The US Army ultimately rejected the robot for being too noisy, according to the Guardian, but since then the company has scaled down its design, first to the Spot, then a first edition of the SpotMini that came out last year.
The latter came with a robotic arm where its head should be and was touted as a domestic helper, but a sleeker second edition without the arm was released earlier this month. There’s little detail on what the new robot is designed for, but the more polished design suggests a more consumer-focused purpose.
OctopusGripper – Festo

Festo has released a long line of animal-inspired machines over the years, from a mechanical kangaroo to robotic butterflies. Its latest creation isn’t a full animal—instead it’s a gripper based on an octopus tentacle that can be attached to the end of a robotic arm.
The pneumatically-powered device is made of soft silicone and features two rows of suction cups on its inner edge. By applying compressed air the tentacle can wrap around a wide variety of differently shaped objects, just like its natural counterpart, and a vacuum can be applied to the larger suction cups to grip the object securely. Because it’s soft, it holds promise for robots required to operate safely in collaboration with humans.
CRAM – University of California, Berkeley

Cockroaches are renowned for their hardiness and ability to disappear down cracks that seem far too small for them. Researchers at UC Berkeley decided these capabilities could be useful for search and rescue missions and so set about experimenting on the insects to find out their secrets.
They found the bugs can squeeze into gaps a fifth of their normal standing height by splaying their legs out to the side without significantly slowing themselves down. So they built a palm-sized robot with a jointed plastic shell that could do the same to squeeze into crevices half its normal height.
Snake Robot – Carnegie Mellon University

Search and rescue missions are a common theme for animal-inspired robots, but the snake robot built by CMU researchers is one of the first to be tested in a real disaster.
A team of roboticists from the university helped Mexican Red Cross workers search collapsed buildings for survivors after the 7.1-magnitude earthquake that struck Mexico City in September. The snake design provides a small diameter and the ability to move in almost any direction, which makes the robot ideal for accessing tight spaces, though the team was unable to locate any survivors.
The snake currently features a camera on the front, but researchers told IEEE Spectrum that the experience helped them realize they should also add a microphone to listen for people trapped under the rubble.
Bio-Hybrid Stingray – Harvard University

Taking more than just inspiration from the animal kingdom, a group from Harvard built a robotic stingray out of silicone and rat heart muscle cells.
The robot uses the same synchronized undulations along the edge of its fins to propel itself as a ray does. But while a ray has two sets of muscles to pull the fins up and down, the new device has only one that pulls them down, with a springy gold skeleton that pulls them back up again. The cells are also genetically modified to be activated by flashes of light.
The project’s leader eventually hopes to engineer a human heart, and both his stingray and an earlier jellyfish bio-robot are primarily aimed at better understanding how that organ works.
Bat Bot – Caltech

Most recent advances in drone technology have come from quadcopters, but Caltech engineers think rigid devices with rapidly spinning propellers are probably not ideal for use in close quarters with humans.
That’s why they turned to soft-winged bats for inspiration. That’s no easy feat, though, considering bats use more than 40 joints with each flap of their wings, so the team had to optimize down to nine joints to avoid it becoming too bulky. The simplified bat can’t ascend yet, but its onboard computer and sensors let it autonomously carry out glides, turns, and dives.
Salto – UC Berkeley

While even the most advanced robots tend to plod around, tree-dwelling animals have the ability to spring from branch to branch to clear obstacles and climb quickly. This could prove invaluable for search and rescue robots by allowing them to quickly traverse disordered rubble.
UC Berkeley engineers turned to the Senegal bush baby for inspiration after determining it scored highest in “vertical jumping agility”—a combination of how high and how frequently an animal can jump. They recreated its ability to get into a super-low crouch that stores energy in its tendons to create a robot that could carry out parkour-style double jumps off walls to quickly gain height.
Pleurobot – École Polytechnique Fédérale de Lausanne

Normally robots are masters of air, land, or sea, but the robotic salamander built by researchers at EPFL can both walk and swim.
Its designers used X-ray videos to carefully study how the amphibians move before using this to build a true-to-life robotic version using 3D printed bones, motorized joints, and a synthetic nervous system made up of electronic circuitry.
The robot’s low center of mass and segmented legs make it great at navigating rough terrain without losing balance, and the ability to swim gives added versatility. They also hope it will help paleontologists gain a better understanding of the movements of the first tetrapods to transition from water to land, which salamanders are the best living analog of.
Eelume – Eelume

A snakelike body isn’t only useful on land—eels are living proof it’s an efficient way to travel underwater, too. Norwegian robotics company Eelume has borrowed these principles to build a robot capable of sub-sea inspection, maintenance, and repair.
The modular design allows operators to put together their own favored configuration of joints and payloads such as sensors and tools. And while an early version of the robot used the same method of locomotion as an eel, the latest version undergoing sea trials has added a variety of thrusters for greater speeds and more maneuverability.
Image Credit: Boston Dynamics / YouTube Continue reading

Posted in Human Robots | Tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , | Leave a comment

#431315 Better Than Smart Speakers? Japan Is ...

While American internet giants are developing speakers, Japanese companies are working on robots and holograms. They all share a common goal: to create the future platform for the Internet of Things (IoT) and smart homes.
Names like Bocco, EMIEW3, Xperia Assistant, and Gatebox may not ring a bell to most outside of Japan, but Sony, Hitachi, Sharp, and Softbank most certainly do. The companies, along with Japanese start-ups, have developed robots, robot concepts, and even holograms like the ones hiding behind the short list of names.
While there are distinct differences between the various systems, they share the potential to act as a remote control for IoT devices and smart homes. It is a very different direction than that taken by companies like Google, Amazon, and Apple, who have so far focused on building IoT speaker systems.
Bocco robot. Image Credit: Yukai Engineering
“Technology companies are pursuing the platform—or smartphone if you will—for IoT. My impression is that Japanese companies—and Japanese consumers—prefer that such a platform should not just be an object, but a companion,” says Kosuke Tatsumi, designer at Yukai Engineering, a startup that has developed the Bocco robot system.
At Hitachi, a spokesperson said that the company’s human symbiotic service robot, EMIEW3, robot is currently in the field, doing proof-of-value tests at customer sites to investigate needs and potential solutions. This could include working as an interactive control system for the Internet of Things:
“EMIEW3 is able to communicate with humans, thus receive instructions, and as it is connected to a robotics IT platform, it is very much capable of interacting with IoT-based systems,” the spokesperson said.
The power of speech is getting feet
Gartner analysis predicts that there will be 8.4 billion internet-connected devices—collectively making up the Internet of Things—by the end of 2017. 5.2 billion of those devices are in the consumer category. By the end of 2020, the number of IoT devices will rise to 12.8 billion—and that is just in the consumer category.
As a child of the 80s, I can vividly remember how fun it was to have separate remote controls for TV, video, and stereo. I can imagine a situation where my internet-connected refrigerator and ditto thermostat, television, and toaster try to work out who I’m talking to and what I want them to do.
Consensus seems to be that speech will be the way to interact with many/most IoT devices. The same goes for a form of virtual assistant functioning as the IoT platform—or remote control. Almost everything else is still an open ballgame, despite an early surge for speaker-based systems, like those from Amazon, Google, and Apple.
Why robots could rule
Famous android creator and robot scientist Dr. Hiroshi Ishiguro sees the interaction between humans and the AI embedded in speakers or robots as central to both approaches. From there, the approaches differ greatly.
Image Credit: Hiroshi Ishiguro Laboratories
“It is about more than the difference of form. Speaking to an Amazon Echo is not a natural kind of interaction for humans. That is part of what we in Japan are creating in many human-like robot systems,” he says. “The human brain is constructed to recognize and interact with humans. This is part of why it makes sense to focus on developing the body for the AI mind as well as the AI mind itself. In a way, you can describe it as the difference between developing an assistant, which could be said to be what many American companies are currently doing, and a companion, which is more the focus here in Japan.”
Another advantage is that robots are more kawaii—a multifaceted Japanese word that can be translated as “cute”—than speakers are. This makes it easy for people to relate to them and forgive them.
“People are more willing to forgive children when they make mistakes, and the same is true with a robot like Bocco, which is designed to look kawaii and childlike,” Kosuke Tatsumi explains.
Japanese robots and holograms with IoT-control capabilities
So, what exactly do these robot and hologram companions look like, what can they do, and who’s making them? Here are seven examples of Japanese companies working to go a step beyond smart speakers with personable robots and holograms.
1. In 2016 Sony’s mobile division demonstrated the Xperia Agent concept robot that recognizes individual users, is voice controlled, and can do things like control your television and receive calls from services like Skype.

2. Sharp launched their Home Assistant at CES 2016. A robot-like, voice-controlled assistant that can to control, among other things, air conditioning units, and televisions. Sharp has also launched a robotic phone called RoBoHon.
3. Gatebox has created a holographic virtual assistant. Evil tongues will say that it is primarily the expression of an otaku (Japanese for nerd) dream of living with a manga heroine. Gatebox is, however, able to control things like lights, TVs, and other systems through API integration. It also provides its owner with weather-related advice like “remember your umbrella, it looks like it will rain later.” Gatebox can be controlled by voice, gesture, or via an app.
4. Hitachi’s EMIEW3 robot is designed to assist people in businesses and public spaces. It is connected to a robot IT-platform via the cloud that acts as a “remote brain.” Hitachi is currently investigating the business use cases for EMIEW3. This could include the role of controlling platform for IoT devices.

5. Softbank’s Pepper robot has been used as a platform to control use of medical IoT devices such as smart thermometers by Avatarion. The company has also developed various in-house systems that enable Pepper to control IoT-devices like a coffee machine. A user simply asks Pepper to brew a cup of coffee, and it starts the coffee machine for you.
6. Yukai Engineering’s Bocco registers when a person (e.g., young child) comes home and acts as a communication center between that person and other members of the household (e.g., parent still at work). The company is working on integrating voice recognition, voice control, and having Bocco control things like the lights and other connected IoT devices.
7. Last year Toyota launched the Kirobo Mini, a companion robot which aims to, among other things, help its owner by suggesting “places to visit, routes for travel, and music to listen to” during the drive.

Today, Japan. Tomorrow…?
One of the key questions is whether this emerging phenomenon is a purely Japanese thing. If the country’s love of robots makes it fundamentally different. Japan is, after all, a country where new units of Softbank’s Pepper robot routinely sell out in minutes and the RoBoHon robot-phone has its own cafe nights in Tokyo.
It is a country where TV introduces you to friendly, helpful robots like Doraemon and Astro Boy. I, on the other hand, first met robots in the shape of Arnold Schwarzenegger’s Terminator and struggled to work out why robots seemed intent on permanently borrowing things like clothes and motorcycles, not to mention why they hated people called Sarah.
However, research suggests that a big part of the reason why Japanese seem to like robots is a combination of exposure and positive experiences that leads to greater acceptance of them. As robots spread to more and more industries—and into our homes—our acceptance of them will grow.
The argument is also backed by a project by Avatarion, which used Softbank’s Nao-robot as a classroom representative for children who were in the hospital.
“What we found was that the other children quickly adapted to interacting with the robot and treating it as the physical representation of the child who was in hospital. They accepted it very quickly,” Thierry Perronnet, General Manager of Avatarion, explains.
His company has also developed solutions where Softbank’s Pepper robot is used as an in-home nurse and controls various medical IoT devices.
If robots end up becoming our preferred method for controlling IoT devices, it is by no means certain that said robots will be coming from Japan.
“I think that the goal for both Japanese and American companies—including the likes of Google, Amazon, Microsoft, and Apple—is to create human-like interaction. For this to happen, technology needs to evolve and adapt to us and how we are used to interacting with others, in other words, have a more human form. Humans’ speed of evolution cannot keep up with technology’s, so it must be the technology that changes,” Dr. Ishiguro says.
Image Credit: Sony Mobile Communications Continue reading

Posted in Human Robots | Tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , | Leave a comment

#431238 AI Is Easy to Fool—Why That Needs to ...

Con artistry is one of the world’s oldest and most innovative professions, and it may soon have a new target. Research suggests artificial intelligence may be uniquely susceptible to tricksters, and as its influence in the modern world grows, attacks against it are likely to become more common.
The root of the problem lies in the fact that artificial intelligence algorithms learn about the world in very different ways than people do, and so slight tweaks to the data fed into these algorithms can throw them off completely while remaining imperceptible to humans.
Much of the research into this area has been conducted on image recognition systems, in particular those relying on deep learning neural networks. These systems are trained by showing them thousands of examples of images of a particular object until they can extract common features that allow them to accurately spot the object in new images.
But the features they extract are not necessarily the same high-level features a human would be looking for, like the word STOP on a sign or a tail on a dog. These systems analyze images at the individual pixel level to detect patterns shared between examples. These patterns can be obscure combinations of pixel values, in small pockets or spread across the image, that would be impossible to discern for a human, but highly accurate at predicting a particular object.

“An attacker can trick the object recognition algorithm into seeing something that isn’t there, without these alterations being obvious to a human.”

What this means is that by identifying these patterns and overlaying them over a different image, an attacker can trick the object recognition algorithm into seeing something that isn’t there, without these alterations being obvious to a human. This kind of manipulation is known as an “adversarial attack.”
Early attempts to trick image recognition systems this way required access to the algorithm’s inner workings to decipher these patterns. But in 2016 researchers demonstrated a “black box” attack that enabled them to trick such a system without knowing its inner workings.
By feeding the system doctored images and seeing how it classified them, they were able to work out what it was focusing on and therefore generate images they knew would fool it. Importantly, the doctored images were not obviously different to human eyes.
These approaches were tested by feeding doctored image data directly into the algorithm, but more recently, similar approaches have been applied in the real world. Last year it was shown that printouts of doctored images that were then photographed on a smartphone successfully tricked an image classification system.
Another group showed that wearing specially designed, psychedelically-colored spectacles could trick a facial recognition system into thinking people were celebrities. In August scientists showed that adding stickers to stop signs in particular configurations could cause a neural net designed to spot them to misclassify the signs.
These last two examples highlight some of the potential nefarious applications for this technology. Getting a self-driving car to miss a stop sign could cause an accident, either for insurance fraud or to do someone harm. If facial recognition becomes increasingly popular for biometric security applications, being able to pose as someone else could be very useful to a con artist.
Unsurprisingly, there are already efforts to counteract the threat of adversarial attacks. In particular, it has been shown that deep neural networks can be trained to detect adversarial images. One study from the Bosch Center for AI demonstrated such a detector, an adversarial attack that fools the detector, and a training regime for the detector that nullifies the attack, hinting at the kind of arms race we are likely to see in the future.
While image recognition systems provide an easy-to-visualize demonstration, they’re not the only machine learning systems at risk. The techniques used to perturb pixel data can be applied to other kinds of data too.

“Bypassing cybersecurity defenses is one of the more worrying and probable near-term applications for this approach.”

Chinese researchers showed that adding specific words to a sentence or misspelling a word can completely throw off machine learning systems designed to analyze what a passage of text is about. Another group demonstrated that garbled sounds played over speakers could make a smartphone running the Google Now voice command system visit a particular web address, which could be used to download malware.
This last example points toward one of the more worrying and probable near-term applications for this approach: bypassing cybersecurity defenses. The industry is increasingly using machine learning and data analytics to identify malware and detect intrusions, but these systems are also highly susceptible to trickery.
At this summer’s DEF CON hacking convention, a security firm demonstrated they could bypass anti-malware AI using a similar approach to the earlier black box attack on the image classifier, but super-powered with an AI of their own.
Their system fed malicious code to the antivirus software and then noted the score it was given. It then used genetic algorithms to iteratively tweak the code until it was able to bypass the defenses while maintaining its function.
All the approaches noted so far are focused on tricking pre-trained machine learning systems, but another approach of major concern to the cybersecurity industry is that of “data poisoning.” This is the idea that introducing false data into a machine learning system’s training set will cause it to start misclassifying things.
This could be particularly challenging for things like anti-malware systems that are constantly being updated to take into account new viruses. A related approach bombards systems with data designed to generate false positives so the defenders recalibrate their systems in a way that then allows the attackers to sneak in.
How likely it is that these approaches will be used in the wild will depend on the potential reward and the sophistication of the attackers. Most of the techniques described above require high levels of domain expertise, but it’s becoming ever easier to access training materials and tools for machine learning.
Simpler versions of machine learning have been at the heart of email spam filters for years, and spammers have developed a host of innovative workarounds to circumvent them. As machine learning and AI increasingly embed themselves in our lives, the rewards for learning how to trick them will likely outweigh the costs.
Image Credit: Nejron Photo / Shutterstock.com Continue reading

Posted in Human Robots | Tagged , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , | Leave a comment