Tag Archives: exhibit
Dr. Been Kim wants to rip open the black box of deep learning.
A senior researcher at Google Brain, Kim specializes in a sort of AI psychology. Like cognitive psychologists before her, she develops various ways to probe the alien minds of artificial neural networks (ANNs), digging into their gory details to better understand the models and their responses to inputs.
The more interpretable ANNs are, the reasoning goes, the easier it is to reveal potential flaws in their reasoning. And if we understand when or why our systems choke, we’ll know when not to use them—a foundation for building responsible AI.
There are already several ways to tap into ANN reasoning, but Kim’s inspiration for unraveling the AI black box came from an entirely different field: cognitive psychology. The field aims to discover fundamental rules of how the human mind—essentially also a tantalizing black box—operates, Kim wrote with her colleagues.
In a new paper uploaded to the pre-publication server arXiv, the team described a way to essentially perform a human cognitive test on ANNs. The test probes how we automatically complete gaps in what we see, so that they form entire objects—for example, perceiving a circle from a bunch of loose dots arranged along a clock face. Psychologist dub this the “law of completion,” a highly influential idea that led to explanations of how our minds generalize data into concepts.
Because deep neural networks in machine vision loosely mimic the structure and connections of the visual cortex, the authors naturally asked: do ANNs also exhibit the law of completion? And what does that tell us about how an AI thinks?
Enter the Germans
The law of completion is part of a series of ideas from Gestalt psychology. Back in the 1920s, long before the advent of modern neuroscience, a group of German experimental psychologists asked: in this chaotic, flashy, unpredictable world, how do we piece together input in a way that leads to meaningful perceptions?
The result is a group of principles known together as the Gestalt effect: that the mind self-organizes to form a global whole. In the more famous words of Gestalt psychologist Kurt Koffka, our perception forms a whole that’s “something else than the sum of its parts.” Not greater than; just different.
Although the theory has its critics, subsequent studies in humans and animals suggest that the law of completion happens on both the cognitive and neuroanatomical level.
Take a look at the drawing below. You immediately “see” a shape that’s actually the negative: a triangle or a square (A and B). Or you further perceive a 3D ball (C), or a snake-like squiggle (D). Your mind fills in blank spots, so that the final perception is more than just the black shapes you’re explicitly given.
Image Credit: Wikimedia Commons contributors, the free media repository.
Neuroscientists now think that the effect comes from how our visual system processes information. Arranged in multiple layers and columns, lower-level neurons—those first to wrangle the data—tend to extract simpler features such as lines or angles. In Gestalt speak, they “see” the parts.
Then, layer by layer, perception becomes more abstract, until higher levels of the visual system directly interpret faces or objects—or things that don’t really exist. That is, the “whole” emerges.
The Experiment Setup
Inspired by these classical experiments, Kim and team developed a protocol to test the Gestalt effect on feed-forward ANNs: one simple, the other, dubbed the “Inception V3,” far more complex and widely used in the machine vision community.
The main idea is similar to the triangle drawings above. First, the team generated three datasets: one set shows complete, ordinary triangles. The second—the “Illusory” set, shows triangles with the edges removed but the corners intact. Thanks to the Gestalt effect, to us humans these generally still look like triangles. The third set also only shows incomplete triangle corners. But here, the corners are randomly rotated so that we can no longer imagine a line connecting them—hence, no more triangle.
To generate a dataset large enough to tease out small effects, the authors changed the background color, image rotation, and other aspects of the dataset. In all, they produced nearly 1,000 images to test their ANNs on.
“At a high level, we compare an ANN’s activation similarities between the three sets of stimuli,” the authors explained. The process is two steps: first, train the AI on complete triangles. Second, test them on the datasets. If the response is more similar between the illusory set and the complete triangle—rather than the randomly rotated set—it should suggest a sort of Gestalt closure effect in the network.
Right off the bat, the team got their answer: yes, ANNs do seem to exhibit the law of closure.
When trained on natural images, the networks better classified the illusory set as triangles than those with randomized connection weights or networks trained on white noise.
When the team dug into the “why,” things got more interesting. The ability to complete an image correlated with the network’s ability to generalize.
Humans subconsciously do this constantly: anything with a handle made out of ceramic, regardless of shape, could easily be a mug. ANNs still struggle to grasp common features—clues that immediately tells us “hey, that’s a mug!” But when they do, it sometimes allows the networks to better generalize.
“What we observe here is that a network that is able to generalize exhibits…more of the closure effect [emphasis theirs], hinting that the closure effect reflects something beyond simply learning features,” the team wrote.
What’s more, remarkably similar to the visual cortex, “higher” levels of the ANNs showed more of the closure effect than lower layers, and—perhaps unsurprisingly—the more layers a network had, the more it exhibited the closure effect.
As the networks learned, their ability to map out objects from fragments also improved. When the team messed around with the brightness and contrast of the images, the AI still learned to see the forest from the trees.
“Our findings suggest that neural networks trained with natural images do exhibit closure,” the team concluded.
That’s not to say that ANNs recapitulate the human brain. As Google’s Deep Dream, an effort to coax AIs into spilling what they’re perceiving, clearly demonstrates, machine vision sees some truly weird stuff.
In contrast, because they’re modeled after the human visual cortex, perhaps it’s not all that surprising that these networks also exhibit higher-level properties inherent to how we process information.
But to Kim and her colleagues, that’s exactly the point.
“The field of psychology has developed useful tools and insights to study human brains– tools that we may be able to borrow to analyze artificial neural networks,” they wrote.
By tweaking these tools to better analyze machine minds, the authors were able to gain insight on how similarly or differently they see the world from us. And that’s the crux: the point isn’t to say that ANNs perceive the world sort of, kind of, maybe similar to humans. It’s to tap into a wealth of cognitive psychology tools, established over decades using human minds, to probe that of ANNs.
“The work here is just one step along a much longer path,” the authors conclude.
“Understanding where humans and neural networks differ will be helpful for research on interpretability by enlightening the fundamental differences between the two interesting species.”
Image Credit: Popova Alena / Shutterstock.com Continue reading
Facial recognition technology has progressed to point where it now interprets emotions in facial expressions. This type of analysis is increasingly used in daily life. For example, companies can use facial recognition software to help with hiring decisions. Other programs scan the faces in crowds to identify threats to public safety.
Unfortunately, this technology struggles to interpret the emotions of black faces. My new study, published last month, shows that emotional analysis technology assigns more negative emotions to black men’s faces than white men’s faces.
This isn’t the first time that facial recognition programs have been shown to be biased. Google labeled black faces as gorillas. Cameras identified Asian faces as blinking. Facial recognition programs struggled to correctly identify gender for people with darker skin.
My work contributes to a growing call to better understand the hidden bias in artificial intelligence software.
To examine the bias in the facial recognition systems that analyze people’s emotions, I used a data set of 400 NBA player photos from the 2016 to 2017 season, because players are similar in their clothing, athleticism, age and gender. Also, since these are professional portraits, the players look at the camera in the picture.
I ran the images through two well-known types of emotional recognition software. Both assigned black players more negative emotional scores on average, no matter how much they smiled.
For example, consider the official NBA pictures of Darren Collison and Gordon Hayward. Both players are smiling, and, according to the facial recognition and analysis program Face++, Darren Collison and Gordon Hayward have similar smile scores—48.7 and 48.1 out of 100, respectively.
Basketball players Darren Collision (left) and Gordon Hayward (right). basketball-reference.com
However, Face++ rates Hayward’s expression as 59.7 percent happy and 0.13 percent angry and Collison’s expression as 39.2 percent happy and 27 percent angry. Collison is viewed as nearly as angry as he is happy and far angrier than Hayward—despite the facial recognition program itself recognizing that both players are smiling.
In contrast, Microsoft’s Face API viewed both men as happy. Still, Collison is viewed as less happy than Hayward, with 98 and 93 percent happiness scores, respectively. Despite his smile, Collison is even scored with a small amount of contempt, whereas Hayward has none.
Across all the NBA pictures, the same pattern emerges. On average, Face++ rates black faces as twice as angry as white faces. Face API scores black faces as three times more contemptuous than white faces. After matching players based on their smiles, both facial analysis programs are still more likely to assign the negative emotions of anger or contempt to black faces.
Stereotyped by AI
My study shows that facial recognition programs exhibit two distinct types of bias.
First, black faces were consistently scored as angrier than white faces for every smile. Face++ showed this type of bias. Second, black faces were always scored as angrier if there was any ambiguity about their facial expression. Face API displayed this type of disparity. Even if black faces are partially smiling, my analysis showed that the systems assumed more negative emotions as compared to their white counterparts with similar expressions. The average emotional scores were much closer across races, but there were still noticeable differences for black and white faces.
This observation aligns with other research, which suggests that black professionals must amplify positive emotions to receive parity in their workplace performance evaluations. Studies show that people perceive black men as more physically threatening than white men, even when they are the same size.
Some researchers argue that facial recognition technology is more objective than humans. But my study suggests that facial recognition reflects the same biases that people have. Black men’s facial expressions are scored with emotions associated with threatening behaviors more often than white men, even when they are smiling. There is good reason to believe that the use of facial recognition could formalize preexisting stereotypes into algorithms, automatically embedding them into everyday life.
Until facial recognition assesses black and white faces similarly, black people may need to exaggerate their positive facial expressions—essentially smile more—to reduce ambiguity and potentially negative interpretations by the technology.
Although innovative, artificial intelligence can perpetrate and exacerbate existing power dynamics, leading to disparate impact across racial/ethnic groups. Some societal accountability is necessary to ensure fairness to all groups because facial recognition, like most artificial intelligence, is often invisible to the people most affected by its decisions.
Lauren Rhue, Assistant Professor of Information Systems and Analytics, Wake Forest University
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Alex_Po / Shutterstock.com Continue reading
Superconductors are among the most bizarre and exciting materials yet discovered. Counterintuitive quantum-mechanical effects mean that, below a critical temperature, they have zero electrical resistance. This property alone is more than enough to spark the imagination.
A current that could flow forever without losing any energy means transmission of power with virtually no losses in the cables. When renewable energy sources start to dominate the grid and high-voltage transmission across continents becomes important to overcome intermittency, lossless cables will result in substantial savings.
What’s more, a superconducting wire carrying a current that never, ever diminishes would act as a perfect store of electrical energy. Unlike batteries, which degrade over time, if the resistance is truly zero, you could return to the superconductor in a billion years and find that same old current flowing through it. Energy could be captured and stored indefinitely!
With no resistance, a huge current could be passed through the superconducting wire and, in turn, produce magnetic fields of incredible power.
You could use them to levitate trains and produce astonishing accelerations, thereby revolutionizing the transport system. You could use them in power plants—replacing conventional methods which spin turbines in magnetic fields to generate electricity—and in quantum computers as the two-level system required for a “qubit,” in which the zeros and ones are replaced by current flowing clockwise or counterclockwise in a superconductor.
Arthur C. Clarke famously said that any sufficiently advanced technology is indistinguishable from magic; superconductors can certainly seem like magical devices. So, why aren’t they busy remaking the world? There’s a problem—that critical temperature.
For all known materials, it’s hundreds of degrees below freezing. Superconductors also have a critical magnetic field; beyond a certain magnetic field strength, they cease to work. There’s a tradeoff: materials with an intrinsically high critical temperature can also often provide the largest magnetic fields when cooled well below that temperature.
This has meant that superconductor applications so far have been limited to situations where you can afford to cool the components of your system to close to absolute zero: in particle accelerators and experimental nuclear fusion reactors, for example.
But even as some aspects of superconductor technology become mature in limited applications, the search for higher temperature superconductors moves on. Many physicists still believe a room-temperature superconductor could exist. Such a discovery would unleash amazing new technologies.
The Quest for Room-Temperature Superconductors
After Heike Kamerlingh Onnes discovered superconductivity by accident while attempting to prove Lord Kelvin’s theory that resistance would increase with decreasing temperature, theorists scrambled to explain the new property in the hope that understanding it might allow for room-temperature superconductors to be synthesized.
They came up with the BCS theory, which explained some of the properties of superconductors. It also predicted that the dream of technologists, a room-temperature superconductor, could not exist; the maximum temperature for superconductivity according to BCS theory was just 30 K.
Then, in the 1980s, the field changed again with the discovery of unconventional, or high-temperature, superconductivity. “High temperature” is still very cold: the highest temperature for superconductivity achieved was -70°C for hydrogen sulphide at extremely high pressures. For normal pressures, -140°C is near the upper limit. Unfortunately, high-temperature superconductors—which require relatively cheap liquid nitrogen, rather than liquid helium, to cool—are mostly brittle ceramics, which are expensive to form into wires and have limited application.
Given the limitations of high-temperature superconductors, researchers continue to believe there’s a better option awaiting discovery—an incredible new material that checks boxes like superconductivity approaching room temperature, affordability, and practicality.
Without a detailed theoretical understanding of how this phenomenon occurs—although incremental progress happens all the time—scientists can occasionally feel like they’re taking educated guesses at materials that might be likely candidates. It’s a little like trying to guess a phone number, but with the periodic table of elements instead of digits.
Yet the prospect remains, in the words of one researcher, tantalizing. A Nobel Prize and potentially changing the world of energy and electricity is not bad for a day’s work.
Some research focuses on cuprates, complex crystals that contain layers of copper and oxygen atoms. Doping cuprates with various different elements, such exotic compounds as mercury barium calcium copper oxide, are amongst the best superconductors known today.
Research also continues into some anomalous but unexplained reports that graphite soaked in water can act as a room-temperature superconductor, but there’s no indication that this could be used for technological applications yet.
In early 2017, as part of the ongoing effort to explore the most extreme and exotic forms of matter we can create on Earth, researchers managed to compress hydrogen into a metal.
The pressure required to do this was more than that at the core of the Earth and thousands of times higher than that at the bottom of the ocean. Some researchers in the field, called condensed-matter physics, doubt that metallic hydrogen was produced at all.
It’s considered possible that metallic hydrogen could be a room-temperature superconductor. But getting the samples to stick around long enough for detailed testing has proved tricky, with the diamonds containing the metallic hydrogen suffering a “catastrophic failure” under the pressure.
Superconductivity—or behavior that strongly resembles it—was also observed in yttrium barium copper oxide (YBCO) at room temperature in 2014. The only catch was that this electron transport lasted for a tiny fraction of a second and required the material to be bombarded with pulsed lasers.
Not very practical, you might say, but tantalizing nonetheless.
Other new materials display enticing properties too. The 2016 Nobel Prize in Physics was awarded for the theoretical work that characterizes topological insulators—materials that exhibit similarly strange quantum behaviors. They can be considered perfect insulators for the bulk of the material but extraordinarily good conductors in a thin layer on the surface.
Microsoft is betting on topological insulators as the key component in their attempt at a quantum computer. They’ve also been considered potentially important components in miniaturized circuitry.
A number of remarkable electronic transport properties have also been observed in new, “2D” structures—like graphene, these are materials synthesized to be as thick as a single atom or molecule. And research continues into how we can utilize the superconductors we’ve already discovered; for example, some teams are trying to develop insulating material that prevents superconducting HVDC cable from overheating.
Room-temperature superconductivity remains as elusive and exciting as it has been for over a century. It is unclear whether a room-temperature superconductor can exist, but the discovery of high-temperature superconductors is a promising indicator that unconventional and highly useful quantum effects may be discovered in completely unexpected materials.
Perhaps in the future—through artificial intelligence simulations or the serendipitous discoveries of a 21st century Kamerlingh Onnes—this little piece of magic could move into the realm of reality.
Image Credit: ktsdesign / Shutterstock.com Continue reading
New planets found in distant corners of the galaxy. Climate models that may improve our understanding of sea level rise. The emergence of new antimalarial drugs. These scientific advances and discoveries have been in the news in recent months.
While representing wildly divergent disciplines, from astronomy to biotechnology, they all have one thing in common: Artificial intelligence played a key role in their scientific discovery.
One of the more recent and famous examples came out of NASA at the end of 2017. The US space agency had announced an eighth planet discovered in the Kepler-90 system. Scientists had trained a neural network—a computer with a “brain” modeled on the human mind—to re-examine data from Kepler, a space-borne telescope with a four-year mission to seek out new life and new civilizations. Or, more precisely, to find habitable planets where life might just exist.
The researchers trained the artificial neural network on a set of 15,000 previously vetted signals until it could identify true planets and false positives 96 percent of the time. It then went to work on weaker signals from nearly 700 star systems with known planets.
The machine detected Kepler 90i—a hot, rocky planet that orbits its sun about every two Earth weeks—through a nearly imperceptible change in brightness captured when a planet passes a star. It also found a sixth Earth-sized planet in the Kepler-80 system.
AI Handles Big Data
The application of AI to science is being driven by three great advances in technology, according to Ross King from the Manchester Institute of Biotechnology at the University of Manchester, leader of a team that developed an artificially intelligent “scientist” called Eve.
Those three advances include much faster computers, big datasets, and improved AI methods, King said. “These advances increasingly give AI superhuman reasoning abilities,” he told Singularity Hub by email.
AI systems can flawlessly remember vast numbers of facts and extract information effortlessly from millions of scientific papers, not to mention exhibit flawless logical reasoning and near-optimal probabilistic reasoning, King says.
AI systems also beat humans when it comes to dealing with huge, diverse amounts of data.
That’s partly what attracted a team of glaciologists to turn to machine learning to untangle the factors involved in how heat from Earth’s interior might influence the ice sheet that blankets Greenland.
Algorithms juggled 22 geologic variables—such as bedrock topography, crustal thickness, magnetic anomalies, rock types, and proximity to features like trenches, ridges, young rifts, and volcanoes—to predict geothermal heat flux under the ice sheet throughout Greenland.
The machine learning model, for example, predicts elevated heat flux upstream of Jakobshavn Glacier, the fastest-moving glacier in the world.
“The major advantage is that we can incorporate so many different types of data,” explains Leigh Stearns, associate professor of geology at Kansas University, whose research takes her to the polar regions to understand how and why Earth’s great ice sheets are changing, questions directly related to future sea level rise.
“All of the other models just rely on one parameter to determine heat flux, but the [machine learning] approach incorporates all of them,” Stearns told Singularity Hub in an email. “Interestingly, we found that there is not just one parameter…that determines the heat flux, but a combination of many factors.”
The research was published last month in Geophysical Research Letters.
Stearns says her team hopes to apply high-powered machine learning to characterize glacier behavior over both short and long-term timescales, thanks to the large amounts of data that she and others have collected over the last 20 years.
Emergence of Robot Scientists
While Stearns sees machine learning as another tool to augment her research, King believes artificial intelligence can play a much bigger role in scientific discoveries in the future.
“I am interested in developing AI systems that autonomously do science—robot scientists,” he said. Such systems, King explained, would automatically originate hypotheses to explain observations, devise experiments to test those hypotheses, physically run the experiments using laboratory robotics, and even interpret the results. The conclusions would then influence the next cycle of hypotheses and experiments.
His AI scientist Eve recently helped researchers discover that triclosan, an ingredient commonly found in toothpaste, could be used as an antimalarial drug against certain strains that have developed a resistance to other common drug therapies. The research was published in the journal Scientific Reports.
Automation using artificial intelligence for drug discovery has become a growing area of research, as the machines can work orders of magnitude faster than any human. AI is also being applied in related areas, such as synthetic biology for the rapid design and manufacture of microorganisms for industrial uses.
King argues that machines are better suited to unravel the complexities of biological systems, with even the most “simple” organisms are host to thousands of genes, proteins, and small molecules that interact in complicated ways.
“Robot scientists and semi-automated AI tools are essential for the future of biology, as there are simply not enough human biologists to do the necessary work,” he said.
Creating Shockwaves in Science
The use of machine learning, neural networks, and other AI methods can often get better results in a fraction of the time it would normally take to crunch data.
For instance, scientists at the National Center for Supercomputing Applications, located at the University of Illinois at Urbana-Champaign, have a deep learning system for the rapid detection and characterization of gravitational waves. Gravitational waves are disturbances in spacetime, emanating from big, high-energy cosmic events, such as the massive explosion of a star known as a supernova. The “Holy Grail” of this type of research is to detect gravitational waves from the Big Bang.
Dubbed Deep Filtering, the method allows real-time processing of data from LIGO, a gravitational wave observatory comprised of two enormous laser interferometers located thousands of miles apart in California and Louisiana. The research was published in Physics Letters B. You can watch a trippy visualization of the results below.
In a more down-to-earth example, scientists published a paper last month in Science Advances on the development of a neural network called ConvNetQuake to detect and locate minor earthquakes from ground motion measurements called seismograms.
ConvNetQuake uncovered 17 times more earthquakes than traditional methods. Scientists say the new method is particularly useful in monitoring small-scale seismic activity, which has become more frequent, possibly due to fracking activities that involve injecting wastewater deep underground. You can learn more about ConvNetQuake in this video:
King says he believes that in the long term there will be no limit to what AI can accomplish in science. He and his team, including Eve, are currently working on developing cancer therapies under a grant from DARPA.
“Robot scientists are getting smarter and smarter; human scientists are not,” he says. “Indeed, there is arguably a case that human scientists are less good. I don’t see any scientist alive today of the stature of a Newton or Einstein—despite the vast number of living scientists. The Physics Nobel [laureate] Frank Wilczek is on record as saying (10 years ago) that in 100 years’ time the best physicist will be a machine. I agree.”
Image Credit: Romaset / Shutterstock.com Continue reading