Forecasting Sequential Data Using Consistent Koopman Autoencoders

Authors: Omri Azencot, N. Benjamin Erichson, Vanessa Lin, Michael Mahoney

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our method on a wide range of high-dimensional and short-term dependent problems, and it achieves accurate estimates for significant prediction horizons, while also being robust to noise.
Researcher Affiliation Academia 1Department of Mathematics at UC Los Angeles, CA, USA. 2ICSI and Department of Statistics at UC Berkeley, CA, USA.
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at github.com/erichson/koopman AE.
Open Datasets Yes We extract a subset of the NOAA OI SST V2 High Resolution Dataset hereafter SST, and we refer to (Reynolds et al., 2007) for additional details.
Dataset Splits No The paper does not explicitly provide details about a validation dataset split. It mentions splitting data into training and test sets but omits specific validation split information.
Hardware Specification No The paper does not provide specific hardware details such as exact GPU/CPU models, processor types, or memory amounts used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers, such as library or solver names and their exact versions.
Experiment Setup Yes Our network minimizes Eq. (13) with a decaying learning rate initially set to 0.01. We fix the loss weights to λid = λfwd = 1, λbwd = 0.1, and λcon = 0.01, for the AE, forward forecast, backward prediction and consistency, respectively. We use λs = 8 prediction steps forward and backward in time.