Reinforcement learning

The last decade has seen vast improvements in humanoid robots, but graduating to widespread use might require going back to the fundamentals.

Neural networks and other forms of machine learning ultimately learn by trial and error, one improvement at a time.

By bringing together disparate approaches, machines can reach a new level of creative problem-solving.

Large language models still struggle with basic reasoning tasks. Two new papers that apply machine learning to math provide a blueprint for how that could change.

A wave of research improves reinforcement learning algorithms by pre-training them as if they were human.

Inspired by the results of a game-playing neural network, mathematicians have been making unexpected advances on an age-old math problem.

Intelligent beings learn by interacting with the world. Artificial intelligence researchers have adopted a similar strategy to teach their virtual agents new tricks.

A new computer program fashioned after artificial intelligence systems like AlphaGo has solved several open problems in combinatorics and graph theory.

Two teams have shown how quantum approaches can solve problems faster than classical computers, bringing physics and computer science closer together.