Conference on Lifelong Learning Agents – CoLLAs 2025

Understanding and Exploiting Embodied Conversation

From August 11 to 14, Rimvydas Rubavicius participated in the 4th Conference on Lifelong Learning Agents (CoLLAs), held at the University of Pennsylvania in Philadelphia. During the conference, Rimvydas had an opportunity to present our recent project, SECURE: Semantics-aware embodied conversation under unawareness for lifelong robot learning, in which we designed an interactive task learning framework for learning to solve tasks under unawareness by exploiting the formal semantic analysis of embodied conversation messages. Such technology is essential for designing intelligent agents that can understand and exploit various learning opportunities that natural interaction may facilitate.  The work done as part of this project was accepted and will be published in the Proceedings of Machine Learning Research (PMLR)

Designing Lifelong Learning Agents

Traditional machine learning has primarily relied on the assumption that data samples are i.i.d., which emphasises optimising test performance within a single task. While this approach has been highly influential, it limits further progress. One of the defining traits of human intelligence is the ability to learn continually and adaptively. In recent years, attention has shifted to alternative paradigms that embrace continual learning under non-i.i.d. and non-stationary settings and move away from a single fixed objective. CoLLAs aims to build a community for researchers in this area.

CoLLAs Website
More about SECURE