Tag Archives: learning

#434837 In Defense of Black Box AI

Deep learning is powering some amazing new capabilities, but we find it hard to scrutinize the workings of these algorithms. Lack of interpretability in AI is a common concern and many are trying to fix it, but is it really always necessary to know what’s going on inside these “black boxes”?

In a recent perspective piece for Science, Elizabeth Holm, a professor of materials science and engineering at Carnegie Mellon University, argued in defense of the black box algorithm. I caught up with her last week to find out more.

Edd Gent: What’s your experience with black box algorithms?

Elizabeth Holm: I got a dual PhD in materials science and engineering and scientific computing. I came to academia about six years ago and part of what I wanted to do in making this career change was to refresh and revitalize my computer science side.

I realized that computer science had changed completely. It used to be about algorithms and making codes run fast, but now it’s about data and artificial intelligence. There are the interpretable methods like random forest algorithms, where we can tell how the machine is making its decisions. And then there are the black box methods, like convolutional neural networks.

Once in a while we can find some information about their inner workings, but most of the time we have to accept their answers and kind of probe around the edges to figure out the space in which we can use them and how reliable and accurate they are.

EG: What made you feel like you had to mount a defense of these black box algorithms?

EH: When I started talking with my colleagues, I found that the black box nature of many of these algorithms was a real problem for them. I could understand that because we’re scientists, we always want to know why and how.

It got me thinking as a bit of a contrarian, “Are black boxes all bad? Must we reject them?” Surely not, because human thought processes are fairly black box. We often rely on human thought processes that the thinker can’t necessarily explain.

It’s looking like we’re going to be stuck with these methods for a while, because they’re really helpful. They do amazing things. And so there’s a very pragmatic realization that these are the best methods we’ve got to do some really important problems, and we’re not right now seeing alternatives that are interpretable. We’re going to have to use them, so we better figure out how.

EG: In what situations do you think we should be using black box algorithms?

EH: I came up with three rules. The simplest rule is: when the cost of a bad decision is small and the value of a good decision is high, it’s worth it. The example I gave in the paper is targeted advertising. If you send an ad no one wants it doesn’t cost a lot. If you’re the receiver it doesn’t cost a lot to get rid of it.

There are cases where the cost is high, and that’s then we choose the black box if it’s the best option to do the job. Things get a little trickier here because we have to ask “what are the costs of bad decisions, and do we really have them fully characterized?” We also have to be very careful knowing that our systems may have biases, they may have limitations in where you can apply them, they may be breakable.

But at the same time, there are certainly domains where we’re going to test these systems so extensively that we know their performance in virtually every situation. And if their performance is better than the other methods, we need to do it. Self driving vehicles are a significant example—it’s almost certain they’re going to have to use black box methods, and that they’re going to end up being better drivers than humans.

The third rule is the more fun one for me as a scientist, and that’s the case where the black box really enlightens us as to a new way to look at something. We have trained a black box to recognize the fracture energy of breaking a piece of metal from a picture of the broken surface. It did a really good job, and humans can’t do this and we don’t know why.

What the computer seems to be seeing is noise. There’s a signal in that noise, and finding it is very difficult, but if we do we may find something significant to the fracture process, and that would be an awesome scientific discovery.

EG: Do you think there’s been too much emphasis on interpretability?

EH: I think the interpretability problem is a fundamental, fascinating computer science grand challenge and there are significant issues where we need to have an interpretable model. But how I would frame it is not that there’s too much emphasis on interpretability, but rather that there’s too much dismissiveness of uninterpretable models.

I think that some of the current social and political issues surrounding some very bad black box outcomes have convinced people that all machine learning and AI should be interpretable because that will somehow solve those problems.

Asking humans to explain their rationale has not eliminated bias, or stereotyping, or bad decision-making in humans. Relying too much on interpreted ability perhaps puts the responsibility in the wrong place for getting better results. I can make a better black box without knowing exactly in what way the first one was bad.

EG: Looking further into the future, do you think there will be situations where humans will have to rely on black box algorithms to solve problems we can’t get our heads around?

EH: I do think so, and it’s not as much of a stretch as we think it is. For example, humans don’t design the circuit map of computer chips anymore. We haven’t for years. It’s not a black box algorithm that designs those circuit boards, but we’ve long since given up trying to understand a particular computer chip’s design.

With the billions of circuits in every computer chip, the human mind can’t encompass it, either in scope or just the pure time that it would take to trace every circuit. There are going to be cases where we want a system so complex that only the patience that computers have and their ability to work in very high-dimensional spaces is going to be able to do it.

So we can continue to argue about interpretability, but we need to acknowledge that we’re going to need to use black boxes. And this is our opportunity to do our due diligence to understand how to use them responsibly, ethically, and with benefits rather than harm. And that’s going to be a social conversation as well as as a scientific one.

*Responses have been edited for length and style

Image Credit: Chingraph / Shutterstock.com Continue reading

Posted in Human Robots

#434818 Watch These Robots Do Tasks You Thought ...

Robots have been masters of manufacturing at speed and precision for decades, but give them a seemingly simple task like stacking shelves, and they quickly get stuck. That’s changing, though, as engineers build systems that can take on the deceptively tricky tasks most humans can do with their eyes closed.

Boston Dynamics is famous for dramatic reveals of robots performing mind-blowing feats that also leave you scratching your head as to what the market is—think the bipedal Atlas doing backflips or Spot the galloping robot dog.

Last week, the company released a video of a robot called Handle that looks like an ostrich on wheels carrying out the seemingly mundane task of stacking boxes in a warehouse.

It might seem like a step backward, but this is exactly the kind of practical task robots have long struggled with. While the speed and precision of industrial robots has seen them take over many functions in modern factories, they’re generally limited to highly prescribed tasks carried out in meticulously-controlled environments.

That’s because despite their mechanical sophistication, most are still surprisingly dumb. They can carry out precision welding on a car or rapidly assemble electronics, but only by rigidly following a prescribed set of motions. Moving cardboard boxes around a warehouse might seem simple to a human, but it actually involves a variety of tasks machines still find pretty difficult—perceiving your surroundings, navigating, and interacting with objects in a dynamic environment.

But the release of this video suggests Boston Dynamics thinks these kinds of applications are close to prime time. Last week the company doubled down by announcing the acquisition of start-up Kinema Systems, which builds computer vision systems for robots working in warehouses.

It’s not the only company making strides in this area. On the same day the video went live, Google unveiled a robot arm called TossingBot that can pick random objects from a box and quickly toss them into another container beyond its reach, which could prove very useful for sorting items in a warehouse. The machine can train on new objects in just an hour or two, and can pick and toss up to 500 items an hour with better accuracy than any of the humans who tried the task.

And an apple-picking robot built by Abundant Robotics is currently on New Zealand farms navigating between rows of apple trees using LIDAR and computer vision to single out ripe apples before using a vacuum tube to suck them off the tree.

In most cases, advances in machine learning and computer vision brought about by the recent AI boom are the keys to these rapidly improving capabilities. Robots have historically had to be painstakingly programmed by humans to solve each new task, but deep learning is making it possible for them to quickly train themselves on a variety of perception, navigation, and dexterity tasks.

It’s not been simple, though, and the application of deep learning in robotics has lagged behind other areas. A major limitation is that the process typically requires huge amounts of training data. That’s fine when you’re dealing with image classification, but when that data needs to be generated by real-world robots it can make the approach impractical. Simulations offer the possibility to run this training faster than real time, but it’s proved difficult to translate policies learned in virtual environments into the real world.

Recent years have seen significant progress on these fronts, though, and the increasing integration of modern machine learning with robotics. In October, OpenAI imbued a robotic hand with human-level dexterity by training an algorithm in a simulation using reinforcement learning before transferring it to the real-world device. The key to ensuring the translation went smoothly was injecting random noise into the simulation to mimic some of the unpredictability of the real world.

And just a couple of weeks ago, MIT researchers demonstrated a new technique that let a robot arm learn to manipulate new objects with far less training data than is usually required. By getting the algorithm to focus on a few key points on the object necessary for picking it up, the system could learn to pick up a previously unseen object after seeing only a few dozen examples (rather than the hundreds or thousands typically required).

How quickly these innovations will trickle down to practical applications remains to be seen, but a number of startups as well as logistics behemoth Amazon are developing robots designed to flexibly pick and place the wide variety of items found in your average warehouse.

Whether the economics of using robots to replace humans at these kinds of menial tasks makes sense yet is still unclear. The collapse of collaborative robotics pioneer Rethink Robotics last year suggests there are still plenty of challenges.

But at the same time, the number of robotic warehouses is expected to leap from 4,000 today to 50,000 by 2025. It may not be long until robots are muscling in on tasks we’ve long assumed only humans could do.

Image Credit: Visual Generation / Shutterstock.com Continue reading

Posted in Human Robots

#434797 This Week’s Awesome Stories From ...

GENE EDITING
Genome Engineers Made More Than 13,000 Genome Edits in a Single Cell
Antonio Regalado | MIT Technology Review
“The group, led by gene technologist George Church, wants to rewrite genomes at a far larger scale than has currently been possible, something it says could ultimately lead to the ‘radical redesign’ of species—even humans.”

ROBOTICS
Inside Google’s Rebooted Robotics Program
Cade Metz | The New York Times
“Google’s new lab is indicative of a broader effort to bring so-called machine learning to robotics. …Many believe that machine learning—not extravagant new devices—will be the key to developing robotics for manufacturing, warehouse automation, transportation and many other tasks.

VIDEOS
Boston Dynamics Builds the Warehouse Robot of Jeff Bezos’ Dreams
Luke Dormehl | Digital Trends
“…for anyone wondering what the future of warehouse operation is likely to look like, this offers a far more practical glimpse of the years to come than, say, a dancing dog robot. As Boston Dynamics moves toward commercializing its creations for the first time, this could turn out to be a lot closer than you might think.”

TECHNOLOGY
Europe Is Splitting the Internet Into Three
Casey Newton | The Verge
“The internet had previously been divided into two: the open web, which most of the world could access; and the authoritarian web of countries like China, which is parceled out stingily and heavily monitored. As of today, though, the web no longer feels truly worldwide. Instead we now have the American internet, the authoritarian internet, and the European internet. How does the EU Copyright Directive change our understanding of the web?”

VIRTUAL REALITY
No Man’s Sky’s Next Update Will Let You Explore Infinite Space in Virtual Reality
Taylor Hatmaker | TechCrunch
“Assuming the game runs well enough, No Man’s Sky Virtual Reality will be a far cry from gimmicky VR games that lack true depth, offering one of the most expansive—if not the most expansive—VR experiences to date.”

3D PRINTING
3D Metal Printing Tries to Break Into the Manufacturing Mainstream
Mark Anderson | IEEE Spectrum
“It’s been five or so years since 3D printing was at peak hype. Since then, the technology has edged its way into a new class of materials and started to break into more applications. Today, 3D printers are being seriously considered as a means to produce stainless steel 5G smartphones, high-strength alloy gas-turbine blades, and other complex metal parts.”

Image Credit: ale de sun / Shutterstock.com Continue reading

Posted in Human Robots

#434786 AI Performed Like a Human on a Gestalt ...

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.

Machine Gestalt
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.

AI Psychology
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

Posted in Human Robots

#434781 What Would It Mean for AI to Become ...

As artificial intelligence systems take on more tasks and solve more problems, it’s hard to say which is rising faster: our interest in them or our fear of them. Futurist Ray Kurzweil famously predicted that “By 2029, computers will have emotional intelligence and be convincing as people.”

We don’t know how accurate this prediction will turn out to be. Even if it takes more than 10 years, though, is it really possible for machines to become conscious? If the machines Kurzweil describes say they’re conscious, does that mean they actually are?

Perhaps a more relevant question at this juncture is: what is consciousness, and how do we replicate it if we don’t understand it?

In a panel discussion at South By Southwest titled “How AI Will Design the Human Future,” experts from academia and industry discussed these questions and more.

Wait, What Is AI?
Most of AI’s recent feats—diagnosing illnesses, participating in debate, writing realistic text—involve machine learning, which uses statistics to find patterns in large datasets then uses those patterns to make predictions. However, “AI” has been used to refer to everything from basic software automation and algorithms to advanced machine learning and deep learning.

“The term ‘artificial intelligence’ is thrown around constantly and often incorrectly,” said Jennifer Strong, a reporter at the Wall Street Journal and host of the podcast “The Future of Everything.” Indeed, one study found that 40 percent of European companies that claim to be working on or using AI don’t actually use it at all.

Dr. Peter Stone, associate chair of computer science at UT Austin, was the study panel chair on the 2016 One Hundred Year Study on Artificial Intelligence (or AI100) report. Based out of Stanford University, AI100 is studying and anticipating how AI will impact our work, our cities, and our lives.

“One of the first things we had to do was define AI,” Stone said. They defined it as a collection of different technologies inspired by the human brain to be able to perceive their surrounding environment and figure out what actions to take given these inputs.

Modeling on the Unknown
Here’s the crazy thing about that definition (and about AI itself): we’re essentially trying to re-create the abilities of the human brain without having anything close to a thorough understanding of how the human brain works.

“We’re starting to pair our brains with computers, but brains don’t understand computers and computers don’t understand brains,” Stone said. Dr. Heather Berlin, cognitive neuroscientist and professor of psychiatry at the Icahn School of Medicine at Mount Sinai, agreed. “It’s still one of the greatest mysteries how this three-pound piece of matter can give us all our subjective experiences, thoughts, and emotions,” she said.

This isn’t to say we’re not making progress; there have been significant neuroscience breakthroughs in recent years. “This has been the stuff of science fiction for a long time, but now there’s active work being done in this area,” said Amir Husain, CEO and founder of Austin-based AI company Spark Cognition.

Advances in brain-machine interfaces show just how much more we understand the brain now than we did even a few years ago. Neural implants are being used to restore communication or movement capabilities in people who’ve been impaired by injury or illness. Scientists have been able to transfer signals from the brain to prosthetic limbs and stimulate specific circuits in the brain to treat conditions like Parkinson’s, PTSD, and depression.

But much of the brain’s inner workings remain a deep, dark mystery—one that will have to be further solved if we’re ever to get from narrow AI, which refers to systems that can perform specific tasks and is where the technology stands today, to artificial general intelligence, or systems that possess the same intelligence level and learning capabilities as humans.

The biggest question that arises here, and one that’s become a popular theme across stories and films, is if machines achieve human-level general intelligence, does that also mean they’d be conscious?

Wait, What Is Consciousness?
As valuable as the knowledge we’ve accumulated about the brain is, it seems like nothing more than a collection of disparate facts when we try to put it all together to understand consciousness.

“If you can replace one neuron with a silicon chip that can do the same function, then replace another neuron, and another—at what point are you still you?” Berlin asked. “These systems will be able to pass the Turing test, so we’re going to need another concept of how to measure consciousness.”

Is consciousness a measurable phenomenon, though? Rather than progressing by degrees or moving through some gray area, isn’t it pretty black and white—a being is either conscious or it isn’t?

This may be an outmoded way of thinking, according to Berlin. “It used to be that only philosophers could study consciousness, but now we can study it from a scientific perspective,” she said. “We can measure changes in neural pathways. It’s subjective, but depends on reportability.”

She described three levels of consciousness: pure subjective experience (“Look, the sky is blue”), awareness of one’s own subjective experience (“Oh, it’s me that’s seeing the blue sky”), and relating one subjective experience to another (“The blue sky reminds me of a blue ocean”).

“These subjective states exist all the way down the animal kingdom. As humans we have a sense of self that gives us another depth to that experience, but it’s not necessary for pure sensation,” Berlin said.

Husain took this definition a few steps farther. “It’s this self-awareness, this idea that I exist separate from everything else and that I can model myself,” he said. “Human brains have a wonderful simulator. They can propose a course of action virtually, in their minds, and see how things play out. The ability to include yourself as an actor means you’re running a computation on the idea of yourself.”

Most of the decisions we make involve envisioning different outcomes, thinking about how each outcome would affect us, and choosing which outcome we’d most prefer.

“Complex tasks you want to achieve in the world are tied to your ability to foresee the future, at least based on some mental model,” Husain said. “With that view, I as an AI practitioner don’t see a problem implementing that type of consciousness.”

Moving Forward Cautiously (But Not too Cautiously)
To be clear, we’re nowhere near machines achieving artificial general intelligence or consciousness, and whether a “conscious machine” is possible—not to mention necessary or desirable—is still very much up for debate.

As machine intelligence continues to advance, though, we’ll need to walk the line between progress and risk management carefully.

Improving the transparency and explainability of AI systems is one crucial goal AI developers and researchers are zeroing in on. Especially in applications that could mean the difference between life and death, AI shouldn’t advance without people being able to trace how it’s making decisions and reaching conclusions.

Medicine is a prime example. “There are already advances that could save lives, but they’re not being used because they’re not trusted by doctors and nurses,” said Stone. “We need to make sure there’s transparency.” Demanding too much transparency would also be a mistake, though, because it will hinder the development of systems that could at best save lives and at worst improve efficiency and free up doctors to have more face time with patients.

Similarly, self-driving cars have great potential to reduce deaths from traffic fatalities. But even though humans cause thousands of deadly crashes every day, we’re terrified by the idea of self-driving cars that are anything less than perfect. “If we only accept autonomous cars when there’s zero probability of an accident, then we will never accept them,” Stone said. “Yet we give 16-year-olds the chance to take a road test with no idea what’s going on in their brains.”

This brings us back to the fact that, in building tech modeled after the human brain—which has evolved over millions of years—we’re working towards an end whose means we don’t fully comprehend, be it something as basic as choosing when to brake or accelerate or something as complex as measuring consciousness.

“We shouldn’t charge ahead and do things just because we can,” Stone said. “The technology can be very powerful, which is exciting, but we have to consider its implications.”

Image Credit: agsandrew / Shutterstock.com Continue reading

Posted in Human Robots