Tag Archives: media
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
One of the most contentious debates in technology is around the question of automation and jobs. At issue is whether advances in automation, specifically with regards to artificial intelligence and robotics, will spell trouble for today’s workers. This debate is played out in the media daily, and passions run deep on both sides of the issue. In the past, however, automation has created jobs and increased real wages.
A widespread concern with the current scenario is that the workers most likely to be displaced by technology lack the skills needed to do the new jobs that same technology will create.
Let’s look at this concern in detail. Those who fear automation will hurt workers start by pointing out that there is a wide range of jobs, from low-pay, low-skill to high-pay, high-skill ones. This can be represented as follows:
They then point out that technology primarily creates high-paying jobs, like geneticists, as shown in the diagram below.
Meanwhile, technology destroys low-wage, low-skill jobs like those in fast food restaurants, as shown below:
Then, those who are worried about this dynamic often pose the question, “Do you really think a fast-food worker is going to become a geneticist?”
They worry that we are about to face a huge amount of systemic permanent unemployment, as the unskilled displaced workers are ill-equipped to do the jobs of tomorrow.
It is important to note that both sides of the debate are in agreement at this point. Unquestionably, technology destroys low-skilled, low-paying jobs while creating high-skilled, high-paying ones.
So, is that the end of the story? As a society are we destined to bifurcate into two groups, those who have training and earn high salaries in the new jobs, and those with less training who see their jobs vanishing to machines? Is this latter group forever locked out of economic plenty because they lack training?
The question, “Can a fast food worker become a geneticist?” is where the error comes in. Fast food workers don’t become geneticists. What happens is that a college biology professor becomes a geneticist. Then a high-school biology teacher gets the college job. Then the substitute teacher gets hired on full-time to fill the high school teaching job. All the way down.
The question is not whether those in the lowest-skilled jobs can do the high-skilled work. Instead the question is, “Can everyone do a job just a little harder than the job they have today?” If so, and I believe very deeply that this is the case, then every time technology creates a new job “at the top,” everyone gets a promotion.
This isn’t just an academic theory—it’s 200 years of economic history in the west. For 200 years, with the exception of the Great Depression, unemployment in the US has been between 2 percent and 13 percent. Always. Europe’s range is a bit wider, but not much.
If I took 200 years of unemployment rates and graphed them, and asked you to find where the assembly line took over manufacturing, or where steam power rapidly replaced animal power, or the lightning-fast adoption of electricity by industry, you wouldn’t be able to find those spots. They aren’t even blips in the unemployment record.
You don’t even have to look back as far as the assembly line to see this happening. It has happened non-stop for 200 years. Every fifty years, we lose about half of all jobs, and this has been pretty steady since 1800.
How is it that for 200 years we have lost half of all jobs every half century, but never has this process caused unemployment? Not only has it not caused unemployment, but during that time, we have had full employment against the backdrop of rising wages.
How can wages rise while half of all jobs are constantly being destroyed? Simple. Because new technology always increases worker productivity. It creates new jobs, like web designer and programmer, while destroying low-wage backbreaking work. When this happens, everyone along the way gets a better job.
Our current situation isn’t any different than the past. The nature of technology has always been to create high-skilled jobs and increase worker productivity. This is good news for everyone.
People often ask me what their children should study to make sure they have a job in the future. I usually say it doesn’t really matter. If I knew everything I know now and went back to the mid 1980s, what could I have taken in high school to make me better prepared for today? There is only one class, and it wasn’t computer science. It was typing. Who would have guessed?
The great skill is to be able to learn new things, and luckily, we all have that. In fact, that is our singular ability as a species. What I do in my day-to-day job consists largely of skills I have learned as the years have passed. In my experience, if you ask people at all job levels,“Would you like a little more challenging job to make a little more money?” almost everyone says yes.
That’s all it has taken for us to collectively get here today, and that’s all we need going forward.
Image Credit: Lightspring / Shutterstock.com Continue reading