Understanding the visual knowledge of language models

LLMs trained primarily on text can generate complex visual concepts through code with self-correction. Researchers used these illustrations to train an image-free computer vision system to recognize real photos.

Alex Shipps | MIT CSAIL • mit
yesterday ~6 min

Technologies enable 3D imaging of whole human brain hemispheres at subcellular resolution

Three innovations by an MIT-based team enable high-resolution, high-throughput imaging of human brain tissue at a full range of scales, and mapping connectivity of neurons at single-cell resolution.

David Orenstein | The Picower Institute for Learning and Memory • mit
yesterday ~12 min

The Hubble telescope has shifted into one-gyro mode after months of technical issues − an aerospace engineering expert explains

Hubble’s technical issues continue. But through some clever engineering, the telescope can continue operations with just 1 gyroscope.

Panagiotis Tsiotras, Professor of Aerospace Engineering, Georgia Institute of Technology • conversation
yesterday ~6 min

A smarter way to streamline drug discovery

The SPARROW algorithm automatically identifies the best molecules to test as potential new medicines, given the vast number of factors affecting each choice.

Adam Zewe | MIT News • mit
yesterday ~8 min

Technique improves the reasoning capabilities of large language models

Combining natural language and programming, the method enables LLMs to solve numerical, analytical, and language-based tasks transparently.

Adam Zewe | MIT News • mit
June 14, 2024 ~7 min

Researchers use large language models to help robots navigate

The method uses language-based inputs instead of costly visual data to direct a robot through a multistep navigation task.

Adam Zewe | MIT News • mit
June 12, 2024 ~7 min

New algorithm discovers language just by watching videos

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.

Rachel Gordon | MIT CSAIL • mit
June 11, 2024 ~9 min

What’s going on in our brains when we plan?

Study uncovers how the brain simulates possible future actions by drawing from our stored memories.

Cambridge University News • cambridge
June 11, 2024 ~5 min

New computer vision method helps speed up screening of electronic materials

The technique characterizes a material’s electronic properties 85 times faster than conventional methods.

Jennifer Chu | MIT News • mit
June 11, 2024 ~8 min

How do you build tunnels and bridges underwater? A geotechnical engineer explains the construction tricks

Underwater construction is a complex and difficult task, but engineers have developed several ways to build underwater … mostly by not building underwater at all.

Ari Perez, Associate Professor of Civil Engineering, Quinnipiac University • conversation
June 10, 2024 ~9 min