Tag Archives: family
#437269 DeepMind’s Newest AI Programs Itself ...
When Deep Blue defeated world chess champion Garry Kasparov in 1997, it may have seemed artificial intelligence had finally arrived. A computer had just taken down one of the top chess players of all time. But it wasn’t to be.
Though Deep Blue was meticulously programmed top-to-bottom to play chess, the approach was too labor-intensive, too dependent on clear rules and bounded possibilities to succeed at more complex games, let alone in the real world. The next revolution would take a decade and a half, when vastly more computing power and data revived machine learning, an old idea in artificial intelligence just waiting for the world to catch up.
Today, machine learning dominates, mostly by way of a family of algorithms called deep learning, while symbolic AI, the dominant approach in Deep Blue’s day, has faded into the background.
Key to deep learning’s success is the fact the algorithms basically write themselves. Given some high-level programming and a dataset, they learn from experience. No engineer anticipates every possibility in code. The algorithms just figure it.
Now, Alphabet’s DeepMind is taking this automation further by developing deep learning algorithms that can handle programming tasks which have been, to date, the sole domain of the world’s top computer scientists (and take them years to write).
In a paper recently published on the pre-print server arXiv, a database for research papers that haven’t been peer reviewed yet, the DeepMind team described a new deep reinforcement learning algorithm that was able to discover its own value function—a critical programming rule in deep reinforcement learning—from scratch.
Surprisingly, the algorithm was also effective beyond the simple environments it trained in, going on to play Atari games—a different, more complicated task—at a level that was, at times, competitive with human-designed algorithms and achieving superhuman levels of play in 14 games.
DeepMind says the approach could accelerate the development of reinforcement learning algorithms and even lead to a shift in focus, where instead of spending years writing the algorithms themselves, researchers work to perfect the environments in which they train.
Pavlov’s Digital Dog
First, a little background.
Three main deep learning approaches are supervised, unsupervised, and reinforcement learning.
The first two consume huge amounts of data (like images or articles), look for patterns in the data, and use those patterns to inform actions (like identifying an image of a cat). To us, this is a pretty alien way to learn about the world. Not only would it be mind-numbingly dull to review millions of cat images, it’d take us years or more to do what these programs do in hours or days. And of course, we can learn what a cat looks like from just a few examples. So why bother?
While supervised and unsupervised deep learning emphasize the machine in machine learning, reinforcement learning is a bit more biological. It actually is the way we learn. Confronted with several possible actions, we predict which will be most rewarding based on experience—weighing the pleasure of eating a chocolate chip cookie against avoiding a cavity and trip to the dentist.
In deep reinforcement learning, algorithms go through a similar process as they take action. In the Atari game Breakout, for instance, a player guides a paddle to bounce a ball at a ceiling of bricks, trying to break as many as possible. When playing Breakout, should an algorithm move the paddle left or right? To decide, it runs a projection—this is the value function—of which direction will maximize the total points, or rewards, it can earn.
Move by move, game by game, an algorithm combines experience and value function to learn which actions bring greater rewards and improves its play, until eventually, it becomes an uncanny Breakout player.
Learning to Learn (Very Meta)
So, a key to deep reinforcement learning is developing a good value function. And that’s difficult. According to the DeepMind team, it takes years of manual research to write the rules guiding algorithmic actions—which is why automating the process is so alluring. Their new Learned Policy Gradient (LPG) algorithm makes solid progress in that direction.
LPG trained in a number of toy environments. Most of these were “gridworlds”—literally two-dimensional grids with objects in some squares. The AI moves square to square and earns points or punishments as it encounters objects. The grids vary in size, and the distribution of objects is either set or random. The training environments offer opportunities to learn fundamental lessons for reinforcement learning algorithms.
Only in LPG’s case, it had no value function to guide that learning.
Instead, LPG has what DeepMind calls a “meta-learner.” You might think of this as an algorithm within an algorithm that, by interacting with its environment, discovers both “what to predict,” thereby forming its version of a value function, and “how to learn from it,” applying its newly discovered value function to each decision it makes in the future.
Prior work in the area has had some success, but according to DeepMind, LPG is the first algorithm to discover reinforcement learning rules from scratch and to generalize beyond training. The latter was particularly surprising because Atari games are so different from the simple worlds LPG trained in—that is, it had never seen anything like an Atari game.
Time to Hand Over the Reins? Not Just Yet
LPG is still behind advanced human-designed algorithms, the researchers said. But it outperformed a human-designed benchmark in training and even some Atari games, which suggests it isn’t strictly worse, just that it specializes in some environments.
This is where there’s room for improvement and more research.
The more environments LPG saw, the more it could successfully generalize. Intriguingly, the researchers speculate that with enough well-designed training environments, the approach might yield a general-purpose reinforcement learning algorithm.
At the least, though, they say further automation of algorithm discovery—that is, algorithms learning to learn—will accelerate the field. In the near term, it can help researchers more quickly develop hand-designed algorithms. Further out, as self-discovered algorithms like LPG improve, engineers may shift from manually developing the algorithms themselves to building the environments where they learn.
Deep learning long ago left Deep Blue in the dust at games. Perhaps algorithms learning to learn will be a winning strategy in the real world too.
Image credit: Mike Szczepanski / Unsplash Continue reading
#436466 How Two Robots Learned to Grill and ...
The list of things robots can do seems to be growing by the week. They can play sports, help us explore outer space and the deep sea, take over some of our boring everyday tasks, and even assemble Ikea furniture.
Now they can add one more accomplishment to the list: grilling and serving a hot dog.
It seems like a pretty straightforward task, and as far as grilling goes, hot dogs are about as easy as it gets (along with, maybe, burgers? Hot dogs require more rotation, but it’s easier to tell when they’re done since they’re lighter in color).
Let’s paint a picture: you’re manning the grill at your family’s annual Fourth of July celebration. You’ve got a 10-pack of plump, juicy beef franks and a hungry crowd of relatives whose food-to-alcohol ratio is getting pretty skewed—they need some solid calories, pronto. What are the steps you need to take to get those franks from package to plate?
Each one needs to be placed on the grill, rotated every couple minutes for even cooking, removed from the grill when you deem it’s done, then—if you’re the kind of guy or gal who goes the extra mile—placed in a bun and dressed with ketchup, mustard, pickles, and the like before being handed over to salivating, too-loud Uncle Hector or sweet, bored Cousin Margaret.
While carrying out your grillmaster duties, you know better than to drop the hot dogs on the ground, leave them cooking on one side for too long, squeeze them to the point of breaking or bursting, and any other hot-dog-ruining amateur moves.
But for a robot, that’s a lot to figure out, especially if they have no prior knowledge of grilling hot dogs (which, well, most robots don’t).
As described in a paper published in this week’s Science Robotics, a team from Boston University programmed two robotic arms to use reinforcement learning—a branch of machine learning in which software gathers information about its environment then learns from it by replaying its experiences and incorporating rewards—to cook and serve hot dogs.
The team used a set of formulas to specify and combine tasks (“pick up hot dog and place on the grill”), meet safety requirements (“always avoid collisions”), and incorporate general prior knowledge (“you cannot pick up another hot dog if you are already holding one”).
Baxter and Jaco—as the two robots were dubbed—were trained through computer simulations. The paper’s authors emphasized their use of what they call a “formal specification language” for training the software, with the aim of generating easily-interpretable task descriptions. In reinforcement learning, they explain, being able to understand how a reward function influences an AI’s learning process is a key component in understanding the system’s behavior—but most systems lack this quality, and are thus likely to be lumped into the ‘black box’ of AI.
The robots’ decisions throughout the hot dog prep process—when to turn a hot dog, when to take it off the grill, and so on—are, the authors write, “easily interpretable from the beginning because the language is very similar to plain English.”
Besides being a step towards more explainable AI systems, Baxter and Jaco are another example of fast-food robots—following in the footsteps of their burger and pizza counterparts—that may take over some repetitive manual tasks currently performed by human workers. As robots’ capabilities improve through incremental progress like this, they’ll be able to take on additional tasks.
In a not-so-distant future, then, you just may find yourself throwing back drinks with Uncle Hector and Cousin Margaret while your robotic replacement mans the grill, churning out hot dogs that are perfectly cooked every time.
Image Credit: Image by Muhammad Ribkhan from Pixabay Continue reading