DenseAV, developed at MIT, learns to parse and understand the meaning of language just by watching videos of people talking, with potential applications in multimedia search, language learning, and robotics.
The technique characterizes a material’s electronic properties 85 times faster than conventional methods.
In the new economics course 14.163 (Algorithms and Behavioral Science), students investigate the deployment of machine-learning tools and their potential to understand people, reduce bias, and improve society.
With generative AI models, researchers combined robotics data from different sources to help robots learn better.
A new approach could streamline virtual training processes or aid clinicians in reviewing diagnostic videos.
“Alchemist” system adjusts the material attributes of specific objects within images to potentially modify video game models to fit different environments, fine-tune VFX, and diversify robotic training.
Fifteen new faculty members join six of the school’s academic departments.
The 10 Design Fellows are MIT graduate students working at the intersection of design and multiple disciplines across the Institute.
A new technique that can automatically classify phases of physical systems could help scientists investigate novel materials.
A new “consensus game,” developed by MIT CSAIL researchers, elevates AI’s text comprehension and generation skills.
