MIT Gives a Century-Old Economic Model a Major Machine Learning Upgrade

MIT researchers gave a major machine learning upgrade to random utility models, a nearly century-old economic framework for predicting choices. Meanwhile a separate MIT finding shows an overlooked algorithm class outperforms expectations in certain games, and a new deep learning approach captures atomic patterns to improve material property predictions. Here’s this week’s machine learning research roundup.

Read more

High Schooler’s ML Algorithm Discovers 1.5 Million Unknown Light Sources

A Pasadena high school student’s machine learning algorithm discovered 1.5 million previously unknown variable light sources in NASA’s NEOWISE data, while researchers warn autonomous AI agents could pose an existential threat to grant funding systems. Here’s what this week’s machine learning research reveals.

Read more