Tag Archives: inspiration

#432467 Dungeons and Dragons, Not Chess and Go: ...

Everyone had died—not that you’d know it, from how they were laughing about their poor choices and bad rolls of the dice. As a social anthropologist, I study how people understand artificial intelligence (AI) and our efforts towards attaining it; I’m also a life-long fan of Dungeons and Dragons (D&D), the inventive fantasy roleplaying game. During a recent quest, when I was playing an elf ranger, the trainee paladin (or holy knight) acted according to his noble character, and announced our presence at the mouth of a dragon’s lair. The results were disastrous. But while success in D&D means “beating the bad guy,” the game is also a creative sandbox, where failure can count as collective triumph so long as you tell a great tale.

What does this have to do with AI? In computer science, games are frequently used as a benchmark for an algorithm’s “intelligence.” The late Robert Wilensky, a professor at the University of California, Berkeley and a leading figure in AI, offered one reason why this might be. Computer scientists “looked around at who the smartest people were, and they were themselves, of course,” he told the authors of Compulsive Technology: Computers as Culture (1985). “They were all essentially mathematicians by training, and mathematicians do two things—they prove theorems and play chess. And they said, hey, if it proves a theorem or plays chess, it must be smart.” No surprise that demonstrations of AI’s “smarts” have focused on the artificial player’s prowess.

Yet the games that get chosen—like Go, the main battlefield for Google DeepMind’s algorithms in recent years—tend to be tightly bounded, with set objectives and clear paths to victory or defeat. These experiences have none of the open-ended collaboration of D&D. Which got me thinking: do we need a new test for intelligence, where the goal is not simply about success, but storytelling? What would it mean for an AI to “pass” as human in a game of D&D? Instead of the Turing test, perhaps we need an elf ranger test?

Of course, this is just a playful thought experiment, but it does highlight the flaws in certain models of intelligence. First, it reveals how intelligence has to work across a variety of environments. D&D participants can inhabit many characters in many games, and the individual player can “switch” between roles (the fighter, the thief, the healer). Meanwhile, AI researchers know that it’s super difficult to get a well-trained algorithm to apply its insights in even slightly different domains—something that we humans manage surprisingly well.

Second, D&D reminds us that intelligence is embodied. In computer games, the bodily aspect of the experience might range from pressing buttons on a controller in order to move an icon or avatar (a ping-pong paddle; a spaceship; an anthropomorphic, eternally hungry, yellow sphere), to more recent and immersive experiences involving virtual-reality goggles and haptic gloves. Even without these add-ons, games can still produce biological responses associated with stress and fear (if you’ve ever played Alien: Isolation you’ll understand). In the original D&D, the players encounter the game while sitting around a table together, feeling the story and its impact. Recent research in cognitive science suggests that bodily interactions are crucial to how we grasp more abstract mental concepts. But we give minimal attention to the embodiment of artificial agents, and how that might affect the way they learn and process information.

Finally, intelligence is social. AI algorithms typically learn through multiple rounds of competition, in which successful strategies get reinforced with rewards. True, it appears that humans also evolved to learn through repetition, reward and reinforcement. But there’s an important collaborative dimension to human intelligence. In the 1930s, the psychologist Lev Vygotsky identified the interaction of an expert and a novice as an example of what became called “scaffolded” learning, where the teacher demonstrates and then supports the learner in acquiring a new skill. In unbounded games, this cooperation is channelled through narrative. Games of It among small children can evolve from win/lose into attacks by terrible monsters, before shifting again to more complex narratives that explain why the monsters are attacking, who is the hero, and what they can do and why—narratives that aren’t always logical or even internally compatible. An AI that could engage in social storytelling is doubtless on a surer, more multifunctional footing than one that plays chess; and there’s no guarantee that chess is even a step on the road to attaining intelligence of this sort.

In some ways, this failure to look at roleplaying as a technical hurdle for intelligence is strange. D&D was a key cultural touchstone for technologists in the 1980s and the inspiration for many early text-based computer games, as Katie Hafner and Matthew Lyon point out in Where Wizards Stay up Late: The Origins of the Internet (1996). Even today, AI researchers who play games in their free time often mention D&D specifically. So instead of beating adversaries in games, we might learn more about intelligence if we tried to teach artificial agents to play together as we do: as paladins and elf rangers.

This article was originally published at Aeon and has been republished under Creative Commons.

Image Credit:Benny Mazur/Flickr / CC BY 2.0 Continue reading

Posted in Human Robots

#431999 Brain-Like Chips Now Beat the Human ...

Move over, deep learning. Neuromorphic computing—the next big thing in artificial intelligence—is on fire.

Just last week, two studies individually unveiled computer chips modeled after information processing in the human brain.

The first, published in Nature Materials, found a perfect solution to deal with unpredictability at synapses—the gap between two neurons that transmit and store information. The second, published in Science Advances, further amped up the system’s computational power, filling synapses with nanoclusters of supermagnetic material to bolster information encoding.

The result? Brain-like hardware systems that compute faster—and more efficiently—than the human brain.

“Ultimately we want a chip as big as a fingernail to replace one big supercomputer,” said Dr. Jeehwan Kim, who led the first study at MIT in Cambridge, Massachusetts.

Experts are hopeful.

“The field’s full of hype, and it’s nice to see quality work presented in an objective way,” said Dr. Carver Mead, an engineer at the California Institute of Technology in Pasadena not involved in the work.

Software to Hardware
The human brain is the ultimate computational wizard. With roughly 100 billion neurons densely packed into the size of a small football, the brain can deftly handle complex computation at lightning speed using very little energy.

AI experts have taken note. The past few years saw brain-inspired algorithms that can identify faces, falsify voices, and play a variety of games at—and often above—human capability.

But software is only part of the equation. Our current computers, with their transistors and binary digital systems, aren’t equipped to run these powerful algorithms.

That’s where neuromorphic computing comes in. The idea is simple: fabricate a computer chip that mimics the brain at the hardware level. Here, data is both processed and stored within the chip in an analog manner. Each artificial synapse can accumulate and integrate small bits of information from multiple sources and fire only when it reaches a threshold—much like its biological counterpart.

Experts believe the speed and efficiency gains will be enormous.

For one, the chips will no longer have to transfer data between the central processing unit (CPU) and storage blocks, which wastes both time and energy. For another, like biological neural networks, neuromorphic devices can support neurons that run millions of streams of parallel computation.

A “Brain-on-a-chip”
Optimism aside, reproducing the biological synapse in hardware form hasn’t been as easy as anticipated.

Neuromorphic chips exist in many forms, but often look like a nanoscale metal sandwich. The “bread” pieces are generally made of conductive plates surrounding a switching medium—a conductive material of sorts that acts like the gap in a biological synapse.

When a voltage is applied, as in the case of data input, ions move within the switching medium, which then creates conductive streams to stimulate the downstream plate. This change in conductivity mimics the way biological neurons change their “weight,” or the strength of connectivity between two adjacent neurons.

But so far, neuromorphic synapses have been rather unpredictable. According to Kim, that’s because the switching medium is often comprised of material that can’t channel ions to exact locations on the downstream plate.

“Once you apply some voltage to represent some data with your artificial neuron, you have to erase and be able to write it again in the exact same way,” explains Kim. “But in an amorphous solid, when you write again, the ions go in different directions because there are lots of defects.”

In his new study, Kim and colleagues swapped the jelly-like switching medium for silicon, a material with only a single line of defects that acts like a channel to guide ions.

The chip starts with a thin wafer of silicon etched with a honeycomb-like pattern. On top is a layer of silicon germanium—something often present in transistors—in the same pattern. This creates a funnel-like dislocation, a kind of Grand Canal that perfectly shuttles ions across the artificial synapse.

The researchers then made a neuromorphic chip containing these synapses and shot an electrical zap through them. Incredibly, the synapses’ response varied by only four percent—much higher than any neuromorphic device made with an amorphous switching medium.

In a computer simulation, the team built a multi-layer artificial neural network using parameters measured from their device. After tens of thousands of training examples, their neural network correctly recognized samples 95 percent of the time, just 2 percent lower than state-of-the-art software algorithms.

The upside? The neuromorphic chip requires much less space than the hardware that runs deep learning algorithms. Forget supercomputers—these chips could one day run complex computations right on our handheld devices.

A Magnetic Boost
Meanwhile, in Boulder, Colorado, Dr. Michael Schneider at the National Institute of Standards and Technology also realized that the standard switching medium had to go.

“There must be a better way to do this, because nature has figured out a better way to do this,” he says.

His solution? Nanoclusters of magnetic manganese.

Schneider’s chip contained two slices of superconducting electrodes made out of niobium, which channel electricity with no resistance. When researchers applied different magnetic fields to the synapse, they could control the alignment of the manganese “filling.”

The switch gave the chip a double boost. For one, by aligning the switching medium, the team could predict the ion flow and boost uniformity. For another, the magnetic manganese itself adds computational power. The chip can now encode data in both the level of electrical input and the direction of the magnetisms without bulking up the synapse.

It seriously worked. At one billion times per second, the chips fired several orders of magnitude faster than human neurons. Plus, the chips required just one ten-thousandth of the energy used by their biological counterparts, all the while synthesizing input from nine different sources in an analog manner.

The Road Ahead
These studies show that we may be nearing a benchmark where artificial synapses match—or even outperform—their human inspiration.

But to Dr. Steven Furber, an expert in neuromorphic computing, we still have a ways before the chips go mainstream.

Many of the special materials used in these chips require specific temperatures, he says. Magnetic manganese chips, for example, require temperatures around absolute zero to operate, meaning they come with the need for giant cooling tanks filled with liquid helium—obviously not practical for everyday use.

Another is scalability. Millions of synapses are necessary before a neuromorphic device can be used to tackle everyday problems such as facial recognition. So far, no deal.

But these problems may in fact be a driving force for the entire field. Intense competition could push teams into exploring different ideas and solutions to similar problems, much like these two studies.

If so, future chips may come in diverse flavors. Similar to our vast array of deep learning algorithms and operating systems, the computer chips of the future may also vary depending on specific requirements and needs.

It is worth developing as many different technological approaches as possible, says Furber, especially as neuroscientists increasingly understand what makes our biological synapses—the ultimate inspiration—so amazingly efficient.

Image Credit: arakio / Shutterstock.com Continue reading

Posted in Human Robots

#431995 The 10 Grand Challenges Facing Robotics ...

Robotics research has been making great strides in recent years, but there are still many hurdles to the machines becoming a ubiquitous presence in our lives. The journal Science Robotics has now identified 10 grand challenges the field will have to grapple with to make that a reality.

Editors conducted an online survey on unsolved challenges in robotics and assembled an expert panel of roboticists to shortlist the 30 most important topics, which were then grouped into 10 grand challenges that could have major impact in the next 5 to 10 years. Here’s what they came up with.

1. New Materials and Fabrication Schemes
Roboticists are beginning to move beyond motors, gears, and sensors by experimenting with things like artificial muscles, soft robotics, and new fabrication methods that combine multiple functions in one material. But most of these advances have been “one-off” demonstrations, which are not easy to combine.

Multi-functional materials merging things like sensing, movement, energy harvesting, or energy storage could allow more efficient robot designs. But combining these various properties in a single machine will require new approaches that blend micro-scale and large-scale fabrication techniques. Another promising direction is materials that can change over time to adapt or heal, but this requires much more research.

2. Bioinspired and Bio-Hybrid Robots
Nature has already solved many of the problems roboticists are trying to tackle, so many are turning to biology for inspiration or even incorporating living systems into their robots. But there are still major bottlenecks in reproducing the mechanical performance of muscle and the ability of biological systems to power themselves.

There has been great progress in artificial muscles, but their robustness, efficiency, and energy and power density need to be improved. Embedding living cells into robots can overcome challenges of powering small robots, as well as exploit biological features like self-healing and embedded sensing, though how to integrate these components is still a major challenge. And while a growing “robo-zoo” is helping tease out nature’s secrets, more work needs to be done on how animals transition between capabilities like flying and swimming to build multimodal platforms.

3. Power and Energy
Energy storage is a major bottleneck for mobile robotics. Rising demand from drones, electric vehicles, and renewable energy is driving progress in battery technology, but the fundamental challenges have remained largely unchanged for years.

That means that in parallel to battery development, there need to be efforts to minimize robots’ power utilization and give them access to new sources of energy. Enabling them to harvest energy from their environment and transmitting power to them wirelessly are two promising approaches worthy of investigation.

4. Robot Swarms
Swarms of simple robots that assemble into different configurations to tackle various tasks can be a cheaper, more flexible alternative to large, task-specific robots. Smaller, cheaper, more powerful hardware that lets simple robots sense their environment and communicate is combining with AI that can model the kind of behavior seen in nature’s flocks.

But there needs to be more work on the most efficient forms of control at different scales—small swarms can be controlled centrally, but larger ones need to be more decentralized. They also need to be made robust and adaptable to the changing conditions of the real world and resilient to deliberate or accidental damage. There also needs to be more work on swarms of non-homogeneous robots with complementary capabilities.

5. Navigation and Exploration
A key use case for robots is exploring places where humans cannot go, such as the deep sea, space, or disaster zones. That means they need to become adept at exploring and navigating unmapped, often highly disordered and hostile environments.

The major challenges include creating systems that can adapt, learn, and recover from navigation failures and are able to make and recognize new discoveries. This will require high levels of autonomy that allow the robots to monitor and reconfigure themselves while being able to build a picture of the world from multiple data sources of varying reliability and accuracy.

6. AI for Robotics
Deep learning has revolutionized machines’ ability to recognize patterns, but that needs to be combined with model-based reasoning to create adaptable robots that can learn on the fly.

Key to this will be creating AI that’s aware of its own limitations and can learn how to learn new things. It will also be important to create systems that are able to learn quickly from limited data rather than the millions of examples used in deep learning. Further advances in our understanding of human intelligence will be essential to solving these problems.

7. Brain-Computer Interfaces
BCIs will enable seamless control of advanced robotic prosthetics but could also prove a faster, more natural way to communicate instructions to robots or simply help them understand human mental states.

Most current approaches to measuring brain activity are expensive and cumbersome, though, so work on compact, low-power, and wireless devices will be important. They also tend to involve extended training, calibration, and adaptation due to the imprecise nature of reading brain activity. And it remains to be seen if they will outperform simpler techniques like eye tracking or reading muscle signals.

8. Social Interaction
If robots are to enter human environments, they will need to learn to deal with humans. But this will be difficult, as we have very few concrete models of human behavior and we are prone to underestimate the complexity of what comes naturally to us.

Social robots will need to be able to perceive minute social cues like facial expression or intonation, understand the cultural and social context they are operating in, and model the mental states of people they interact with to tailor their dealings with them, both in the short term and as they develop long-standing relationships with them.

9. Medical Robotics
Medicine is one of the areas where robots could have significant impact in the near future. Devices that augment a surgeon’s capabilities are already in regular use, but the challenge will be to increase the autonomy of these systems in such a high-stakes environment.

Autonomous robot assistants will need to be able to recognize human anatomy in a variety of contexts and be able to use situational awareness and spoken commands to understand what’s required of them. In surgery, autonomous robots could perform the routine steps of a procedure, giving way to the surgeon for more complicated patient-specific bits.

Micro-robots that operate inside the human body also hold promise, but there are still many roadblocks to their adoption, including effective delivery systems, tracking and control methods, and crucially, finding therapies where they improve on current approaches.

10. Robot Ethics and Security
As the preceding challenges are overcome and robots are increasingly integrated into our lives, this progress will create new ethical conundrums. Most importantly, we may become over-reliant on robots.

That could lead to humans losing certain skills and capabilities, making us unable to take the reins in the case of failures. We may end up delegating tasks that should, for ethical reasons, have some human supervision, and allow people to pass the buck to autonomous systems in the case of failure. It could also reduce self-determination, as human behaviors change to accommodate the routines and restrictions required for robots and AI to work effectively.

Image Credit: Zenzen / Shutterstock.com Continue reading

Posted in Human Robots

#431906 Low-Cost Soft Robot Muscles Can Lift 200 ...

Jerky mechanical robots are staples of science fiction, but to seamlessly integrate into everyday life they’ll need the precise yet powerful motor control of humans. Now scientists have created a new class of artificial muscles that could soon make that a reality.
The advance is the latest breakthrough in the field of soft robotics. Scientists are increasingly designing robots using soft materials that more closely resemble biological systems, which can be more adaptable and better suited to working in close proximity to humans.
One of the main challenges has been creating soft components that match the power and control of the rigid actuators that drive mechanical robots—things like motors and pistons. Now researchers at the University of Colorado Boulder have built a series of low-cost artificial muscles—as little as 10 cents per device—using soft plastic pouches filled with electrically insulating liquids that contract with the force and speed of mammalian skeletal muscles when a voltage is applied to them.

Three different designs of the so-called hydraulically amplified self-healing electrostatic (HASEL) actuators were detailed in two papers in the journals Science and Science Robotics last week. They could carry out a variety of tasks, from gently picking up delicate objects like eggs or raspberries to lifting objects many times their own weight, such as a gallon of water, at rapid repetition rates.
“We draw our inspiration from the astonishing capabilities of biological muscle,” Christoph Keplinger, an assistant professor at UC Boulder and senior author of both papers, said in a press release. “Just like biological muscle, HASEL actuators can reproduce the adaptability of an octopus arm, the speed of a hummingbird and the strength of an elephant.”
The artificial muscles work by applying a voltage to hydrogel electrodes on either side of pouches filled with liquid insulators, which can be as simple as canola oil. This creates an attraction between the two electrodes, pulling them together and displacing the liquid. This causes a change of shape that can push or pull levers, arms or any other articulated component.
The design is essentially a synthesis of two leading approaches to actuating soft robots. Pneumatic and hydraulic actuators that pump fluids around have been popular due to their high forces, easy fabrication and ability to mimic a variety of natural motions. But they tend to be bulky and relatively slow.
Dielectric elastomer actuators apply an electric field across a solid insulating layer to make it flex. These can mimic the responsiveness of biological muscle. But they are not very versatile and can also fail catastrophically, because the high voltages required can cause a bolt of electricity to blast through the insulator, destroying it. The likelihood of this happening increases in line with the size of their electrodes, which makes it hard to scale them up. By combining the two approaches, researchers get the best of both worlds, with the power, versatility and easy fabrication of a fluid-based system and the responsiveness of electrically-powered actuators.
One of the designs holds particular promise for robotics applications, as it behaves a lot like biological muscle. The so-called Peano-HASEL actuators are made up of multiple rectangular pouches connected in series, which allows them to contract linearly, just like real muscle. They can lift more than 200 times their weight, but being electrically powered, they exceed the flexing speed of human muscle.
As the name suggests, the HASEL actuators are also self-healing. They are still prone to the same kind of electrical damage as dielectric elastomer actuators, but the liquid insulator is able to immediately self-heal by redistributing itself and regaining its insulating properties.
The muscles can even monitor the amount of strain they’re under to provide the same kind of feedback biological systems would. The muscle’s capacitance—its ability to store an electric charge—changes as the device stretches, which makes it possible to power the arm while simultaneously measuring what position it’s in.
The researchers say this could imbue robots with a similar sense of proprioception or body-awareness to that found in plants and animals. “Self-sensing allows for the development of closed-loop feedback controllers to design highly advanced and precise robots for diverse applications,” Shane Mitchell, a PhD student in Keplinger’s lab and an author on both papers, said in an email.
The researchers say the high voltages required are an ongoing challenge, though they’ve already designed devices in the lab that use a fifth of the voltage of those features in the recent papers.
In most of their demonstrations, these soft actuators were being used to power rigid arms and levers, pointing to the fact that future robots are likely to combine both rigid and soft components, much like animals do. The potential applications for the technology range from more realistic prosthetics to much more dextrous robots that can work easily alongside humans.
It will take some work before these devices appear in commercial robots. But the combination of high-performance with simple and inexpensive fabrication methods mean other researchers are likely to jump in, so innovation could be rapid.
Image Credit: Keplinger Research Group/University of Colorado Continue reading

Posted in Human Robots

#431836 Do Our Brains Use Deep Learning to Make ...

The first time Dr. Blake Richards heard about deep learning, he was convinced that he wasn’t just looking at a technique that would revolutionize artificial intelligence. He also knew he was looking at something fundamental about the human brain.
That was the early 2000s, and Richards was taking a course with Dr. Geoff Hinton at the University of Toronto. Hinton, a pioneer architect of the algorithm that would later take the world by storm, was offering an introductory course on his learning method inspired by the human brain.
The key words here are “inspired by.” Despite Richards’ conviction, the odds were stacked against him. The human brain, as it happens, seems to lack a critical function that’s programmed into deep learning algorithms. On the surface, the algorithms were violating basic biological facts already proven by neuroscientists.
But what if, superficial differences aside, deep learning and the brain are actually compatible?
Now, in a new study published in eLife, Richards, working with DeepMind, proposed a new algorithm based on the biological structure of neurons in the neocortex. Also known as the cortex, this outermost region of the brain is home to higher cognitive functions such as reasoning, prediction, and flexible thought.
The team networked their artificial neurons together into a multi-layered network and challenged it with a classic computer vision task—identifying hand-written numbers.
The new algorithm performed well. But the kicker is that it analyzed the learning examples in a way that’s characteristic of deep learning algorithms, even though it was completely based on the brain’s fundamental biology.
“Deep learning is possible in a biological framework,” concludes the team.
Because the model is only a computer simulation at this point, Richards hopes to pass the baton to experimental neuroscientists, who could actively test whether the algorithm operates in an actual brain.
If so, the data could then be passed back to computer scientists to work out the next generation of massively parallel and low-energy algorithms to power our machines.
It’s a first step towards merging the two fields back into a “virtuous circle” of discovery and innovation.
The blame game
While you’ve probably heard of deep learning’s recent wins against humans in the game of Go, you might not know the nitty-gritty behind the algorithm’s operations.
In a nutshell, deep learning relies on an artificial neural network with virtual “neurons.” Like a towering skyscraper, the network is structured into hierarchies: lower-level neurons process aspects of an input—for example, a horizontal or vertical stroke that eventually forms the number four—whereas higher-level neurons extract more abstract aspects of the number four.
To teach the network, you give it examples of what you’re looking for. The signal propagates forward in the network (like climbing up a building), where each neuron works to fish out something fundamental about the number four.
Like children trying to learn a skill the first time, initially the network doesn’t do so well. It spits out what it thinks a universal number four should look like—think a Picasso-esque rendition.
But here’s where the learning occurs: the algorithm compares the output with the ideal output, and computes the difference between the two (dubbed “error”). This error is then “backpropagated” throughout the entire network, telling each neuron: hey, this is how far off you were, so try adjusting your computation closer to the ideal.
Millions of examples and tweakings later, the network inches closer to the desired output and becomes highly proficient at the trained task.
This error signal is crucial for learning. Without efficient “backprop,” the network doesn’t know which of its neurons are off kilter. By assigning blame, the AI can better itself.
The brain does this too. How? We have no clue.
Biological No-Go
What’s clear, though, is that the deep learning solution doesn’t work.
Backprop is a pretty needy function. It requires a very specific infrastructure for it to work as expected.
For one, each neuron in the network has to receive the error feedback. But in the brain, neurons are only connected to a few downstream partners (if that). For backprop to work in the brain, early-level neurons need to be able to receive information from billions of connections in their downstream circuits—a biological impossibility.
And while certain deep learning algorithms adapt a more local form of backprop— essentially between neurons—it requires their connection forwards and backwards to be symmetric. This hardly ever occurs in the brain’s synapses.
More recent algorithms adapt a slightly different strategy, in that they implement a separate feedback pathway that helps the neurons to figure out errors locally. While it’s more biologically plausible, the brain doesn’t have a separate computational network dedicated to the blame game.
What it does have are neurons with intricate structures, unlike the uniform “balls” that are currently applied in deep learning.
Branching Networks
The team took inspiration from pyramidal cells that populate the human cortex.
“Most of these neurons are shaped like trees, with ‘roots’ deep in the brain and ‘branches’ close to the surface,” says Richards. “What’s interesting is that these roots receive a different set of inputs than the branches that are way up at the top of the tree.”
This is an illustration of a multi-compartment neural network model for deep learning. Left: Reconstruction of pyramidal neurons from mouse primary visual cortex. Right: Illustration of simplified pyramidal neuron models. Image Credit: CIFAR
Curiously, the structure of neurons often turn out be “just right” for efficiently cracking a computational problem. Take the processing of sensations: the bottoms of pyramidal neurons are right smack where they need to be to receive sensory input, whereas the tops are conveniently placed to transmit feedback errors.
Could this intricate structure be evolution’s solution to channeling the error signal?
The team set up a multi-layered neural network based on previous algorithms. But rather than having uniform neurons, they gave those in middle layers—sandwiched between the input and output—compartments, just like real neurons.
When trained with hand-written digits, the algorithm performed much better than a single-layered network, despite lacking a way to perform classical backprop. The cell-like structure itself was sufficient to assign error: the error signals at one end of the neuron are naturally kept separate from input at the other end.
Then, at the right moment, the neuron brings both sources of information together to find the best solution.
There’s some biological evidence for this: neuroscientists have long known that the neuron’s input branches perform local computations, which can be integrated with signals that propagate backwards from the so-called output branch.
However, we don’t yet know if this is the brain’s way of dealing blame—a question that Richards urges neuroscientists to test out.
What’s more, the network parsed the problem in a way eerily similar to traditional deep learning algorithms: it took advantage of its multi-layered structure to extract progressively more abstract “ideas” about each number.
“[This is] the hallmark of deep learning,” the authors explain.
The Deep Learning Brain
Without doubt, there will be more twists and turns to the story as computer scientists incorporate more biological details into AI algorithms.
One aspect that Richards and team are already eyeing is a top-down predictive function, in which signals from higher levels directly influence how lower levels respond to input.
Feedback from upper levels doesn’t just provide error signals; it could also be nudging lower processing neurons towards a “better” activity pattern in real-time, says Richards.
The network doesn’t yet outperform other non-biologically derived (but “brain-inspired”) deep networks. But that’s not the point.
“Deep learning has had a huge impact on AI, but, to date, its impact on neuroscience has been limited,” the authors say.
Now neuroscientists have a lead they could experimentally test: that the structure of neurons underlie nature’s own deep learning algorithm.
“What we might see in the next decade or so is a real virtuous cycle of research between neuroscience and AI, where neuroscience discoveries help us to develop new AI and AI can help us interpret and understand our experimental data in neuroscience,” says Richards.
Image Credit: christitzeimaging.com / Shutterstock.com Continue reading

Posted in Human Robots