TP6: Frontiers of Machine Learning – Lifelong Learning


  • Davide Bacciu (Università di Pisa)
  • Concetto Spampinato (Università di Catania)

In recent years, artificial intelligence (AI) and machine learning have become pivotal in fields like visual understanding, natural language processing, strategic planning, decision-making, and autonomous robotics. Despite the progress, current approaches still face challenges such as the need for large annotated datasets, limited autonomous learning capabilities, and difficulties in adapting to new environments. These limitations contrast with human and animal learning skills, prompting a need to reformulate AI training paradigms. In this context, the concept of lifelong learning stands out as a pivotal catalyst for pushing the boundaries of intelligence. Despite its potential, current research in this domain is in its early stages, often concentrating on specific issues like class-incremental learning. To confront the pressing challenges of today’s AI landscape, TP6 takes a proactive stance. It aims to navigate contemporary hurdles by advancing continual, incremental, weakly-supervised learning, and meta-learning for AI agents and robots. The primary goal is to enable a seamless post-deployment evolution, effectively addressing challenges such as catastrophic forgetting, ensuring adaptability, and improving generalization across diverse tasks and domains.

This initiative aligns seamlessly with recent breakthroughs in AI, as it delves into innovative perspectives to foster ongoing learning in the ever-evolving realm of artificial intelligence.