New kernel method learns Koopman embeddings, rivals ANNs

Researchers have developed a new kernel-based method for learning finite-dimensional Koopman embeddings directly from data, called extended dynamic mode decomposition with kernel-based dictionary learning (EDMD-kDL). This approach combines bilevel optimization and collocation methods to simultaneously learn a kernel dictionary and a Koopman approximation. EDMD-kDL demonstrates performance comparable to or exceeding state-of-the-art artificial neural network-based methods in various numerical experiments, including sea-surface-temperature forecasting and video data analysis. The method is designed to be scalable to large datasets. AI

IMPACT This new kernel-based method offers an interpretable and scalable alternative to ANNs for learning Koopman embeddings, potentially advancing data-driven modeling of complex systems.

RANK_REASON The cluster contains an academic paper detailing a new method for learning Koopman embeddings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• artificial neural network
• EDMD-kDL
• Extended dynamic mode decomposition with dictionary learning: A data-driven adaptive spectral decomposition of the Koopman operator.
• Joel-Pascal Ntwali N'konzi
• Koopman
• sea surface temperature

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →