Professor Yannakakis recently represented the GenV Project at the 8th annual Artificial Intelligence and Games Summer School, held this year at Leiden University in the Netherlands—an institution ranked among the world's top 100 by Times Higher Education. The prestigious event, which has previously traveled to global hubs like New York, London, Copenhagen, and Malta, serves as a premier forum for high-level knowledge exchange, convening leading academics and industry professionals from powerhouses such as Google DeepMind, Microsoft Research, and Riot Games. Professor Yannakakis showcased the forefront of AI innovation by introducing InvPatch alongside other advanced research topics, fostering valuable cross-sector collaboration and driving dialogue on the future of AI technology.

Researchers at the Institute of Digital Games, University of Malta, presented their paper at IEEE CAI 2026 for their AI framework, InvPatch, which enables AI models to predict the sequence of causal actions from observed videos without relying on rule-based systems or video metadata. Developed by Nemanja Rašajski, Konstantinos Makantasis, Antonios Liapis, and Georgios N. Yannakakis, InvPatch maps visual features into a single global vector to condition the action-prediction process, demonstrating remarkable data efficiency by maintaining comparable performance even when trained with 30% less data. This breakthrough enhances data-frugal inverse dynamics modeling across both synthetic and real-world benchmarks, including a 3% improvement over state-of-the-art results on the KIT Bimanual Actions dataset.