Reducing the complexity of a powerful machine-learning model may help level the playing field for automatic speech-recognition around the world.

A new method forces a machine learning model to focus on more data when learning a task, which leads to more reliable predictions.

This new generation of GelSight’s mobile device offers a sleek form-factor that is one-third lighter and less than half the volume of its predecessor, allowing it to scan surfaces in tighter spaces, while maintaining accuracy, speed, and field of view.

Self-driving Roboats, developed at MIT, set sea in Amsterdam canals. If you don’t get seasick, an autonomous boat might be the right mode of transportation for you. 

Neuroscientists find the internal workings of next-word prediction models resemble those of language-processing centers in the brain.

A new machine-learning system costs less, generates less waste, and can be more innovative than manual discovery methods.

A new control system, demonstrated using MIT’s robotic mini cheetah, enables four-legged robots to jump across uneven terrain in real-time.

Neural Networks can learn to solve all sorts of problems, from identifying cats in photographs to steering a self-driving car. But whether these powerful, pattern-recognizing algorithms actually understand the tasks they are performing remains an open question.

Although the idea of utilizing computers to interpret images is not new, the MIT-led group is drawing on an underused resource—the vast body of radiology reports that accompany medical images, written by radiologists in routine clinical practice—in order to improve the interpretive abilities of machine learning algorithms.