Lifelong learning from event-based data
Document Type
Conference Item
Publication Date
1-1-2021
Abstract
Lifelong learning is a long-standing aim for artificial agents that act in dynamic environments, in which an agent needs to accumulate knowledge incrementally without forgetting previously learned representations. We investigate methods for learning from data produced by event cameras and compare techniques to mitigate forgetting while learning incrementally. We propose a model that is composed of both, feature extraction and continuous learning. Furthermore, we introduce a habituation-based method to mitigate forgetting. Our experimental results show that the combination of different techniques can help to avoid catastrophic forgetting while learning incrementally from the features provided by the extraction module. © 2021 ESANN Intelligence and Machine Learning. All rights reserved.
Keywords
Machine learning, Artificial agents, Catastrophic forgetting, Continuous learning, Dynamic environments, Event-based, Features extraction, Life long learning, Extraction
Divisions
ai
Funders
Alexander von Humboldt-Stiftung,Deutsche Forschungsgemeinschaft [Grant No: TRR169]
Publication Title
ESANN 2021 Proceedings - 29th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Event Title
ESANN 2021 Proceedings - 29th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Event Location
Virtual, Online
Event Dates
6-8 October 2021
Event Type
conference