Tag Archives: train

#439674 Cerebras Upgrades Trillion-Transistor ...

Posted on August 28, 2021 by Android

Much of the recent progress in AI has come from building ever-larger neural networks. A new chip powerful enough to handle “brain-scale” models could turbo-charge this approach.

Chip startup Cerebras leaped into the limelight in 2019 when it came out of stealth to reveal a 1.2-trillion-transistor chip. The size of a dinner plate, the chip is called the Wafer Scale Engine and was the world’s largest computer chip. Earlier this year Cerebras unveiled the Wafer Scale Engine 2 (WSE-2), which more than doubled the number of transistors to 2.6 trillion.

Now the company has outlined a series of innovations that mean its latest chip can train a neural network with up to 120 trillion parameters. For reference, OpenAI’s revolutionary GPT-3 language model contains 175 billion parameters. The largest neural network to date, which was trained by Google, had 1.6 trillion.

“Larger networks, such as GPT-3, have already transformed the natural language processing landscape, making possible what was previously unimaginable,” said Cerebras CEO and co-founder Andrew Feldman in a press release.

“The industry is moving past 1 trillion parameter models, and we are extending that boundary by two orders of magnitude, enabling brain-scale neural networks with 120 trillion parameters.”

The genius of Cerebras’ approach is that rather than taking a silicon wafer and splitting it up to make hundreds of smaller chips, it makes a single massive one. While your average GPU will have a few hundred cores, the WSE-2 has 850,000. Because they’re all on the same hunk of silicon, they can work together far more seamlessly.

This makes the chip ideal for tasks that require huge numbers of operations to happen in parallel, which includes both deep learning and various supercomputing applications. And earlier this week at the Hotchips conference, the company unveiled new technology that is pushing the WSE-2’s capabilities even further.

A major challenge for large neural networks is shuttling around all the data involved in their calculations. Most chips have a limited amount of memory on-chip, and every time data has to be shuffled in and out it creates a bottleneck, which limits the practical size of networks.

The WSE-2 already has an enormous 40 gigabytes of on-chip memory, which means it can hold even the largest of today’s networks. But the company has also built an external unit called MemoryX that provides up to 2.4 Petabytes of high-performance memory, which is so tightly integrated it behaves as if it were on-chip.

Cerebras has also revamped its approach to that data it shuffles around. Previously the guts of the neural network would be stored on the chip, and only the training data would be fed in. Now, though, the weights of the connections between the network’s neurons are kept in the MemoryX unit and streamed in during training.

By combining these two innovations, the company says, they can train networks two orders of magnitude larger than anything that exists today. Other advances announced at the same time include the ability to run extremely sparse (and therefore efficient) neural networks, and a new communication system dubbed SwarmX that makes it possible to link up to 192 chips to create a combined total of 163 million cores.

How much all this cutting-edge technology will cost and who is in a position to take advantage of it is unclear. “This is highly specialized stuff,” Mike Demler, a senior analyst with the Linley Group, told Wired. “It only makes sense for training the very largest models.”

While the size of AI models has been increasing rapidly, it’s likely to be years before anyone can push the WSE-2 to its limits. And despite the insinuations in Cerebras’ press material, just because the parameter count roughly matches the number of synapses in the brain, that doesn’t mean the new chip will be able to run models anywhere close to its complexity or performance.

There’s a major debate in AI circles today over whether we can achieve general artificial intelligence by simply building larger neural networks, or this will require new theoretical breakthroughs. So far, increasing parameter counts has led to pretty consistent jumps in performance. A two-order-of-magnitude improvement over today’s largest models would undoubtedly be significant.

It’s still far from clear whether that trend will hold out, but Cerebras’ new chip could get us considerably closer to an answer.

Image Credit: Cerebras Continue reading →

Posted in Human Robots

#439110 Robotic Exoskeletons Could One Day Walk ...

Posted on April 13, 2021 by Android

Engineers, using artificial intelligence and wearable cameras, now aim to help robotic exoskeletons walk by themselves.

Increasingly, researchers around the world are developing lower-body exoskeletons to help people walk. These are essentially walking robots users can strap to their legs to help them move.

One problem with such exoskeletons: They often depend on manual controls to switch from one mode of locomotion to another, such as from sitting to standing, or standing to walking, or walking on the ground to walking up or down stairs. Relying on joysticks or smartphone apps every time you want to switch the way you want to move can prove awkward and mentally taxing, says Brokoslaw Laschowski, a robotics researcher at the University of Waterloo in Canada.

Scientists are working on automated ways to help exoskeletons recognize when to switch locomotion modes — for instance, using sensors attached to legs that can detect bioelectric signals sent from your brain to your muscles telling them to move. However, this approach comes with a number of challenges, such as how how skin conductivity can change as a person’s skin gets sweatier or dries off.

Now several research groups are experimenting with a new approach: fitting exoskeleton users with wearable cameras to provide the machines with vision data that will let them operate autonomously. Artificial intelligence (AI) software can analyze this data to recognize stairs, doors, and other features of the surrounding environment and calculate how best to respond.

Laschowski leads the ExoNet project, the first open-source database of high-resolution wearable camera images of human locomotion scenarios. It holds more than 5.6 million images of indoor and outdoor real-world walking environments. The team used this data to train deep-learning algorithms; their convolutional neural networks can already automatically recognize different walking environments with 73 percent accuracy “despite the large variance in different surfaces and objects sensed by the wearable camera,” Laschowski notes.

According to Laschowski, a potential limitation of their work their reliance on conventional 2-D images, whereas depth cameras could also capture potentially useful distance data. He and his collaborators ultimately chose not to rely on depth cameras for a number of reasons, including the fact that the accuracy of depth measurements typically degrades in outdoor lighting and with increasing distance, he says.

In similar work, researchers in North Carolina had volunteers with cameras either mounted on their eyeglasses or strapped onto their knees walk through a variety of indoor and outdoor settings to capture the kind of image data exoskeletons might use to see the world around them. The aim? “To automate motion,” says Edgar Lobaton an electrical engineering researcher at North Carolina State University. He says they are focusing on how AI software might reduce uncertainty due to factors such as motion blur or overexposed images “to ensure safe operation. We want to ensure that we can really rely on the vision and AI portion before integrating it into the hardware.”

In the future, Laschowski and his colleagues will focus on improving the accuracy of their environmental analysis software with low computational and memory storage requirements, which are important for onboard, real-time operations on robotic exoskeletons. Lobaton and his team also seek to account for uncertainty introduced into their visual systems by movements .

Ultimately, the ExoNet researchers want to explore how AI software can transmit commands to exoskeletons so they can perform tasks such as climbing stairs or avoiding obstacles based on a system’s analysis of a user's current movements and the upcoming terrain. With autonomous cars as inspiration, they are seeking to develop autonomous exoskeletons that can handle the walking task without human input, Laschowski says.

However, Laschowski adds, “User safety is of the utmost importance, especially considering that we're working with individuals with mobility impairments,” resulting perhaps from advanced age or physical disabilities.
“The exoskeleton user will always have the ability to override the system should the classification algorithm or controller make a wrong decision.” Continue reading →

Posted in Human Robots

#439105 This Robot Taught Itself to Walk in a ...

Posted on April 11, 2021 by Android

Recently, in a Berkeley lab, a robot called Cassie taught itself to walk, a little like a toddler might. Through trial and error, it learned to move in a simulated world. Then its handlers sent it strolling through a minefield of real-world tests to see how it’d fare.

And, as it turns out, it fared pretty damn well. With no further fine-tuning, the robot—which is basically just a pair of legs—was able to walk in all directions, squat down while walking, right itself when pushed off balance, and adjust to different kinds of surfaces.

It’s the first time a machine learning approach known as reinforcement learning has been so successfully applied in two-legged robots.

This likely isn’t the first robot video you’ve seen, nor the most polished.

For years, the internet has been enthralled by videos of robots doing far more than walking and regaining their balance. All that is table stakes these days. Boston Dynamics, the heavyweight champ of robot videos, regularly releases mind-blowing footage of robots doing parkour, back flips, and complex dance routines. At times, it can seem the world of iRobot is just around the corner.

This sense of awe is well-earned. Boston Dynamics is one of the world’s top makers of advanced robots.

But they still have to meticulously hand program and choreograph the movements of the robots in their videos. This is a powerful approach, and the Boston Dynamics team has done incredible things with it.

In real-world situations, however, robots need to be robust and resilient. They need to regularly deal with the unexpected, and no amount of choreography will do. Which is how, it’s hoped, machine learning can help.

Reinforcement learning has been most famously exploited by Alphabet’s DeepMind to train algorithms that thrash humans at some the most difficult games. Simplistically, it’s modeled on the way we learn. Touch the stove, get burned, don’t touch the damn thing again; say please, get a jelly bean, politely ask for another.

In Cassie’s case, the Berkeley team used reinforcement learning to train an algorithm to walk in a simulation. It’s not the first AI to learn to walk in this manner. But going from simulation to the real world doesn’t always translate.

Subtle differences between the two can (literally) trip up a fledgling robot as it tries out its sim skills for the first time.

To overcome this challenge, the researchers used two simulations instead of one. The first simulation, an open source training environment called MuJoCo, was where the algorithm drew upon a large library of possible movements and, through trial and error, learned to apply them. The second simulation, called Matlab SimMechanics, served as a low-stakes testing ground that more precisely matched real-world conditions.

Once the algorithm was good enough, it graduated to Cassie.

And amazingly, it didn’t need further polishing. Said another way, when it was born into the physical world—it knew how to walk just fine. In addition, it was also quite robust. The researchers write that two motors in Cassie’s knee malfunctioned during the experiment, but the robot was able to adjust and keep on trucking.

Other labs have been hard at work applying machine learning to robotics.

Last year Google used reinforcement learning to train a (simpler) four-legged robot. And OpenAI has used it with robotic arms. Boston Dynamics, too, will likely explore ways to augment their robots with machine learning. New approaches—like this one aimed at training multi-skilled robots or this one offering continuous learning beyond training—may also move the dial. It’s early yet, however, and there’s no telling when machine learning will exceed more traditional methods.

And in the meantime, Boston Dynamics bots are testing the commercial waters.

Still, robotics researchers, who were not part of the Berkeley team, think the approach is promising. Edward Johns, head of Imperial College London’s Robot Learning Lab, told MIT Technology Review, “This is one of the most successful examples I have seen.”

The Berkeley team hopes to build on that success by trying out “more dynamic and agile behaviors.” So, might a self-taught parkour-Cassie be headed our way? We’ll see.

Image Credit: University of California Berkeley Hybrid Robotics via YouTube Continue reading →

Posted in Human Robots

#438801 This AI Thrashes the Hardest Atari Games ...

Posted on March 4, 2021 by Android

Learning from rewards seems like the simplest thing. I make coffee, I sip coffee, I’m happy. My brain registers “brewing coffee” as an action that leads to a reward.

That’s the guiding insight behind deep reinforcement learning, a family of algorithms that famously smashed most of Atari’s gaming catalog and triumphed over humans in strategy games like Go. Here, an AI “agent” explores the game, trying out different actions and registering ones that let it win.

Except it’s not that simple. “Brewing coffee” isn’t one action; it’s a series of actions spanning several minutes, where you’re only rewarded at the very end. By just tasting the final product, how do you learn to fine-tune grind coarseness, water to coffee ratio, brewing temperature, and a gazillion other factors that result in the reward—tasty, perk-me-up coffee?

That’s the problem with “sparse rewards,” which are ironically very abundant in our messy, complex world. We don’t immediately get feedback from our actions—no video-game-style dings or points for just grinding coffee beans—yet somehow we’re able to learn and perform an entire sequence of arm and hand movements while half-asleep.

This week, researchers from UberAI and OpenAI teamed up to bestow this talent on AI.

The trick is to encourage AI agents to “return” to a previous step, one that’s promising for a winning solution. The agent then keeps a record of that state, reloads it, and branches out again to intentionally explore other solutions that may have been left behind on the first go-around. Video gamers are likely familiar with this idea: live, die, reload a saved point, try something else, repeat for a perfect run-through.

The new family of algorithms, appropriately dubbed “Go-Explore,” smashed notoriously difficult Atari games like Montezuma’s Revenge that were previously unsolvable by its AI predecessors, while trouncing human performance along the way.

It’s not just games and digital fun. In a computer simulation of a robotic arm, the team found that installing Go-Explore as its “brain” allowed it to solve a challenging series of actions when given very sparse rewards. Because the overarching idea is so simple, the authors say, it can be adapted and expanded to other real-world problems, such as drug design or language learning.

Growing Pains
How do you reward an algorithm?

Rewards are very hard to craft, the authors say. Take the problem of asking a robot to go to a fridge. A sparse reward will only give the robot “happy points” if it reaches its destination, which is similar to asking a baby, with no concept of space and danger, to crawl through a potential minefield of toys and other obstacles towards a fridge.

“In practice, reinforcement learning works very well, if you have very rich feedback, if you can tell, ‘hey, this move is good, that move is bad, this move is good, that move is bad,’” said study author Joost Huinzinga. However, in situations that offer very little feedback, “rewards can intentionally lead to a dead end. Randomly exploring the space just doesn’t cut it.”

The other extreme is providing denser rewards. In the same robot-to-fridge example, you could frequently reward the bot as it goes along its journey, essentially helping “map out” the exact recipe to success. But that’s troubling as well. Over-holding an AI’s hand could result in an extremely rigid robot that ignores new additions to its path—a pet, for example—leading to dangerous situations. It’s a deceptive AI solution that seems effective in a simple environment, but crashes in the real world.

What we need are AI agents that can tackle both problems, the team said.

Intelligent Exploration
The key is to return to the past.

For AI, motivation usually comes from “exploring new or unusual situations,” said Huizinga. It’s efficient, but comes with significant downsides. For one, the AI agent could prematurely stop going back to promising areas because it thinks it had already found a good solution. For another, it could simply forget a previous decision point because of the mechanics of how it probes the next step in a problem.

For a complex task, the end result is an AI that randomly stumbles around towards a solution while ignoring potentially better ones.

“Detaching from a place that was previously visited after collecting a reward doesn’t work in difficult games, because you might leave out important clues,” Huinzinga explained.

Go-Explore solves these problems with a simple principle: first return, then explore. In essence, the algorithm saves different approaches it previously tried and loads promising save points—once more likely to lead to victory—to explore further.

Digging a bit deeper, the AI stores screen caps from a game. It then analyzes saved points and groups images that look alike as a potential promising “save point” to return to. Rinse and repeat. The AI tries to maximize its final score in the game, and updates its save points when it achieves a new record score. Because Atari doesn’t usually allow people to revisit any random point, the team used an emulator, which is a kind of software that mimics the Atari system but with custom abilities such as saving and reloading at any time.

The trick worked like magic. When pitted against 55 Atari games in the OpenAI gym, now commonly used to benchmark reinforcement learning algorithms, Go-Explore knocked out state-of-the-art AI competitors over 85 percent of the time.

It also crushed games previously unbeatable by AI. Montezuma’s Revenge, for example, requires you to move Pedro, the blocky protagonist, through a labyrinth of underground temples while evading obstacles such as traps and enemies and gathering jewels. One bad jump could derail the path to the next level. It’s a perfect example of sparse rewards: you need a series of good actions to get to the reward—advancing onward.

Go-Explore didn’t just beat all levels of the game, a first for AI. It also scored higher than any previous record for reinforcement learning algorithms at lower levels while toppling the human world record.

Outside a gaming environment, Go-Explore was also able to boost the performance of a simulated robot arm. While it’s easy for humans to follow high-level guidance like “put the cup on this shelf in a cupboard,” robots often need explicit training—from grasping the cup to recognizing a cupboard, moving towards it while avoiding obstacles, and learning motions to not smash the cup when putting it down.

Here, similar to the real world, the digital robot arm was only rewarded when it placed the cup onto the correct shelf, out of four possible shelves. When pitted against another algorithm, Go-Explore quickly figured out the movements needed to place the cup, while its competitor struggled with even reliably picking the cup up.

Combining Forces
By itself, the “first return, then explore” idea behind Go-Explore is already powerful. The team thinks it can do even better.

One idea is to change the mechanics of save points. Rather than reloading saved states through the emulator, it’s possible to train a neural network to do the same, without needing to relaunch a saved state. It’s a potential way to make the AI even smarter, the team said, because it can “learn” to overcome one obstacle once, instead of solving the same problem again and again. The downside? It’s much more computationally intensive.

Another idea is to combine Go-Explore with an alternative form of learning, called “imitation learning.” Here, an AI observes human behavior and mimics it through a series of actions. Combined with Go-Explore, said study author Adrien Ecoffet, this could make more robust robots capable of handling all the complexity and messiness in the real world.

To the team, the implications go far beyond Go-Explore. The concept of “first return, then explore” seems to be especially powerful, suggesting “it may be a fundamental feature of learning in general.” The team said, “Harnessing these insights…may be essential…to create generally intelligent agents.”

Image Credit: Adrien Ecoffet, Joost Huizinga, Joel Lehman, Kenneth O. Stanley, and Jeff Clune Continue reading →

Posted in Human Robots

#438774 The World’s First 3D Printed School ...

Posted on February 27, 2021 by Android

3D printed houses have been popping up all over the map. Some are hive-shaped, some can float, some are up for sale. Now this practical, cost-cutting technology is being employed for another type of building: a school.

Located on the island of Madagascar, the project is a collaboration between San Francisco-based architecture firm Studio Mortazavi and Thinking Huts, a nonprofit whose mission is to increase global access to education through 3D printing. The school will be built on the campus of a university in Fianarantsoa, a city in the south central area of the island nation.

According to the World Economic Forum, lack of physical infrastructure is one of the biggest barriers to education. Building schools requires not only funds, human capital, and building materials, but also community collaboration and ongoing upkeep and maintenance. For people to feel good about sending their kids to school each day, the buildings should be conveniently located, appealing, comfortable to spend several hours in, and of course safe. All of this is harder to accomplish than you might think, especially in low-income areas.

Because of its comparatively low cost and quick turnaround time, 3D printing has been lauded as a possible solution to housing shortages and a tool to aid in disaster relief. Cost details of the Madagascar school haven’t been released, but if 3D printed houses can go up in a day for under $10,000 or list at a much lower price than their non-3D-printed neighbors, it’s safe to say that 3D printing a school is likely substantially cheaper than building it through traditional construction methods.

The school’s modular design resembles a honeycomb, where as few or as many nodes as needed can be linked together. Each node consists of a room with two bathrooms, a closet, and a front and rear entrance. The Fianarantsoa school with just have one node to start with, but as local technologists will participate in the building process, they’ll learn the 3D printing ins and outs and subsequently be able to add new nodes or build similar schools in other areas.

Artist rendering of the completed school. Image Credit: Studio Mortazavi/Thinking Huts
The printer for the project is coming from Hyperion Robotics, a Finnish company that specializes in 3D printing solutions for reinforced concrete. The building’s walls will be made of layers of a special cement mixture that Thinking Huts says emits less carbon dioxide than traditional concrete. The roof, doors, and windows will be sourced locally, and the whole process can be completed in less than a week, another major advantage over traditional building methods.

“We can build these schools in less than a week, including the foundation and all the electrical and plumbing work that’s involved,” said Amir Mortazavi, lead architect on the project. “Something like this would typically take months, if not even longer.”

The roof of the building will be equipped with solar panels to provide the school with power, and in a true melding of modern technology and traditional design, the pattern of its walls is based on Malagasy textiles.

Thinking Huts considered seven different countries for its first school, and ended up choosing Madagascar for the pilot based on its need for education infrastructure, stable political outlook, opportunity for growth, and renewable energy potential. However, the team is hoping the pilot will be the first of many similar projects across multiple countries. “We can use this as a case study,” Mortazavi said. “Then we can go to other countries around the world and train the local technologists to use the 3D printer and start a nonprofit there to be able to build schools.”

Construction of the school will take place in the latter half of this year, with hopes of getting students into the classroom as soon as the pandemic is no longer a major threat to the local community’s health.

Image Credit: Studio Mortazavi/Thinking Huts Continue reading →

Posted in Human Robots