Generative Time-series Modeling with Fourier Flows

Authors: Ahmed Alaa, Alex James Chan, Mihaela van der Schaar

ICLR 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments show that Fourier flows perform competitively compared to state-of-the-art baselines.
Researcher Affiliation Academia Ahmed M. Alaa University of California, Los Angeles, USA ahmedmalaa@ucla.edu Alex J. Chan University of Cambridge, UK ajc340@cam.ac.uk Mihaela van der Schaar University of Cambridge, UK University of California, Los Angeles, USA Cambridge Center for AI in Medicine, UK The Alan Turing Institute, UK mv472@cam.ac.uk
Pseudocode No The paper does not contain a clearly labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code No The paper does not include an unambiguous statement about releasing code or a link to a code repository for the described methodology.
Open Datasets Yes We conduct experiments on: Google stocks data, UCI Energy data set, and a longitudinal follow-up clinical data set for lung cancer patients.
Dataset Splits No The paper states using 1,000 synthetic time-series 'to train all baselines' but does not specify explicit train/validation/test splits. For real data, it mentions replicating the experimental setup from Yoon et al. (2019) but does not provide the specific split percentages or methodology within this paper.
Hardware Specification No The paper does not provide specific hardware details such as GPU models, CPU types, or cloud computing instance specifications used for running the experiments.
Software Dependencies No The paper mentions machine learning models (e.g., Bi RNN) but does not specify version numbers for any software dependencies or libraries used in the implementation.
Experiment Setup Yes We train all models with 1,000 iterations and a batch size of 128 we then generate 1,000 synthetic time-series from each trained model. ... a Fourier flow (FF) model with a composition of 10 flows and a (single-layer) Bi RNN with 200 hidden units