Information Sciences Institute

USC computer scientists present a better way to measure the performance of generative AI models at the International Conference on Machine Learning (ICML).

An ISI team identifies limitations in dynamics prediction and proposes a solution.

USC researchers will present nine papers at the 40th International Conference on Machine Learning (ICML 2023), the leading international conference on machine learning.

USC’s Information Science Institute (ISI) brings a host of research to the conference – from dialogue models to machine learning to semantic understanding and more.

Researchers dive into the study of teacher-student natural language interactions to achieve a shared goal in the fantasy world game, Dungeons & Dragons.

USC Information Sciences Institute researchers dive into the nuances of human behavior and how cultural markers affect decision-making.

USC researchers have developed a novel AI technique to pull information from vast amounts of globally sourced text.

A collaborative effort with UCLA will predict psychological health risk factors

USC researchers have proposed a novel method to give personalities to language models.

Russell sees AI from an anthropologist’s perspective – looking at human behavior in an era of machines, with an expertise in collective intelligence and cross-disciplinary thinking.