Two-shot learning of multiple strange attractors

Loading...
Thumbnail Image
Files
1-s2.0-S0893608026006702-main.pdf(5.24 MB)
Published Version
1-s2.0-S0893608026006702-mmc1.pdf(13.01 MB)
Appendix. Supplementary Materials
Date
2026-06-11
Authors
Köglmayr, Daniel
Spahic, Miralem
Flynn, Andrew
Räth, Christoph
Journal Title
Journal ISSN
Volume Title
Publisher
Elsevier Ltd
Research Projects
Organizational Units
Journal Issue
Abstract
The brain combines short- and long-term memory to process, store, and recall multiple different pieces of information. Inspired by this and recent results on multifunctional and parameter-aware learning, we extend a new machine learning technique that combines short- and long-term memory units, specifically, a system consisting of a next-generation reservoir computer (NGRC) and extremely randomized trees (ERT), to process, store, and recall multiple different strange attractors. We train the combined NGRC+ERT system using a two-shot learning approach which significantly improves performance by filtering out unnecessary features, thereby avoiding extensive hyperparameter optimization. We first show that an NGRC+ERT system achieves highly accurate reconstruction of the short- and long-term dynamics of both the Lorenz and Halvorsen chaotic attractors when using an exponential filtering scheme. We validate these findings by training the NGRC+ERT system to reconstruct 16 different attractors and show that sufficient index-based separation in feature space suppresses unwanted switching dynamics, thus stabilizing long-term memory recall. Finally, we identify that defects in short-term memory processing can provoke failure modes in long-term memory recall resulting in confabulation.
Description
Keywords
Attractor reconstruction , Chaotic systems , Extremely randomized trees , Machine learning , Multifunctionality , Reservoir computing
Citation
Köglmayr, D, Spahic, M, Flynn, A & Räth, C 2026, 'Two-shot learning of multiple strange attractors', Neural Networks, vol. 204, 109209, pp. 1-14. https://doi.org/10.1016/j.neunet.2026.109209
Link to publisher’s version
Collections