Training machines to learn more like humans do
Researchers identify a property that helps computer vision models learn to represent the visual world in a more stable, predictable way.
Researchers identify a property that helps computer vision models learn to represent the visual world in a more stable, predictable way.
One of MIT’s five Climate Grand Challenges flagship projects, the Jameel Observatory-CREWSnet project will pilot in Bangladesh and Sudan to help communities adapt to the impacts of climate change.
MIT engineers’ new technique analyzes the 3D organization of the genome at a resolution 100 times higher than before.
The system they developed eliminates a source of bias in simulations, leading to improved algorithms that can boost the performance of applications.
Faulty versions of the Foxp2 gene disrupt neurons’ ability to form synapses in brain regions involved in speech, a new study shows.
A collaborative research team from the MIT-Takeda Program combined physics and machine learning to characterize rough particle surfaces in pharmaceutical pills and powders.
Earth will meet a similar fate in 5 billion years.
SoftZoo is a soft robot co-design platform that can test optimal shapes and sizes for robotic performance in different environments.
Following cataract removal, some of the brain’s visual pathways seem to be more malleable than previously thought.
The research center will support two nonprofits and four government agencies in designing randomized evaluations on housing stability, procedural justice, transportation, income assistance, and more.
Mobility-related data show the pandemic has had a lasting effect, limiting the breadth of places people visit in cities.
The event was spotted in infrared data — also a first — suggesting further searches in this band could turn up more such bursts.
A new method could provide detailed information about internal structures, voids, and cracks, based solely on data about exterior conditions.
California blackworms tangle themselves up by the thousands, then separate in a split second. Their trick may inspire the design of self-detangling materials and fibers.
A new low-temperature growth and fabrication technology allows the integration of 2D materials directly onto a silicon circuit, which could lead to denser and more powerful chips.