Newsroom

In a data-abundant world, the challenge of discerning meaningful patterns becomes ever more crucial. One of the four key re­search questions of STRUC­TURES is how to reliably and efficiently uncover structures in large datasets. Our team of researchers trains specialized unsupervised neural networks to tackle the complexity of structure discovery. These AI algorithms autonomously construct what we term as feature spaces — an abstraction that comes to life through a relatable example: How do people's handwriting styles differ? Our networks possess the ability to learn and interpret distinct handwriting styles. The foundational concepts underlying these abilities can be applied to various structure discovery tasks in diverse fields, including disciplines such as medicine.

Video link (YouTube):
STRUCTURES: How can we efficiently and reliably identify structures in large datasets?