To discover new physics, AI may need to 'unlearn' the old one
A study in the Journal of Cosmology and Astroparticle Physics explores how a machine-learning strategy known as transfer learning could dramatically reduce the computational cost of searching for new physics beyond the standard cosmological model—while also revealing an unexpected risk: Sometimes AI...
Image: Phys.org
A study in the Journal of Cosmology and Astroparticle Physics explores how a machine-learning strategy known as transfer learning could dramatically reduce the computational cost of searching for new physics beyond the standard cosmological model—while also revealing an unexpected risk: Sometimes AI systems can become too reliant on what they already know.
Originally published at Phys.org
