详情

原文传递 Super Turing Evolving Lifelong Learning Architecture (STELLAR).

题名：	Super Turing Evolving Lifelong Learning Architecture (STELLAR).
作者：	Pilly, P. K.
摘要：	The overarching goal of HRL’s Super Turing Lifelong Learning ARchitecture (STELLAR) project was to build a general-purposescalable autonomous system that continually improves its performance during deployment, and rapidly and safely adapts tonew tasks and circumstances, without catastrophic forgetting or the complete undoing of previously learned tasks. Humanbrains are the best available examples of lifelong learners. Leveraging HRL Team’s long-standing expertise in neuroscience,cognitive architectures, machine learning, neuroevolution, and robotics, STELLAR integrates several brain-inspiredmechanisms that operate across multiple spatial and temporal scales. These include memory consolidation or stabilization ofmemories during sleep, neuromodulation, which regulates neural activities and synaptic plasticity, adult neurogenesis or thebirth of neurons, instincts, and reflexes. During Phase 1, we developed nine innovative components that solve differentchallenges and requirements of lifelong learning. During Phase 2, we integrated these components into a complete lifelonglearning system and performed rigorous testing and evaluation. In particular, we demonstrated the efficacy of the fullyintegrated STELLAR system for the autonomous driving domain in the CARLA simulator with online adaptation to differentweather conditions, different vehicle models (wheel asymmetries, vehicle dynamics), and different driving tasks (driving in thecorrect lane vs. opposite lane in regular traffic). The STELLAR system achieved significant performance improvements relativeto conventional machine learning Single Task Experts, exceeding key program targets: about 25% performance improvement,about 7.5x increase in learning efficiency, and about 3.5x improvement in Forward Transfer (leveraging prior learning tofacilitate learning a new task).
总页数：	253 pages