Loss Terms and Operator Forms of Koopman Autoencoders
arXiv:2412.04578v1
Originality Synthesis-oriented
AI Analysis
This work provides a standardized framework for researchers in operator learning, though it is incremental as it builds on existing Koopman autoencoder architectures.
The paper tackles the inconsistency in loss functions and operator forms in Koopman autoencoders by conducting a fair and systematic study, and introduces novel loss terms to address this issue.
Koopman autoencoders are a prevalent architecture in operator learning. But, the loss functions and the form of the operator vary significantly in the literature. This paper presents a fair and systemic study of these options. Furthermore, it introduces novel loss terms.