Do We Need an Encoder-Decoder to Model Dynamical Systems on Networks?

Authors: Bing Liu, Wei Luo, Gang Li, Jing Huang, Bo Yang

IJCAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through extensive experiments, we verify that the proposed model can reliably recover a broad class of dynamics on different network topologies from time series data.
Researcher Affiliation Academia 1College of Computer Science and Technology, Jilin University, China 2Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, China 3School of Information Technology, Deakin University, Geelong, Australia
Pseudocode No The paper describes the model equations and training process but does not include a clearly labeled pseudocode or algorithm block.
Open Source Code No The paper mentions accessing the code for the baseline NDCN model ('https: //github.com/calvin-zcx/ndcn') but does not provide concrete access or an explicit statement about the availability of the source code for their own proposed Dy-Net Neural Dynamics (DNND) model.
Open Datasets No The paper describes simulating its own dataset based on heat equations and other ODEs ('We simulate a vector time series using a heat equation defined on a grid network.' and 'From initial conditions, we simulate a time series as the observation data.'), but does not provide concrete access information (link, DOI, repository, or formal citation) for this generated dataset to be publicly available or open.
Dataset Splits No The paper mentions training data ('randomly sample 80 (irregularly spaced) times 0 t1 < t2 < < t80 5 for training.') and evaluation periods ('evaluated at random times in three periods: interpolation [0-5], short-term [5-6], and long-term [40-50]') which serve as testing, but does not explicitly provide information on a separate validation dataset split.
Hardware Specification No The paper does not provide specific hardware details (e.g., 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, such as library or solver names with version numbers, needed to replicate the experiment.
Experiment Setup Yes We created 80 irregularly spaced samples with t [0, 5], and fitted an NDCN model using the code provided by the authors on Git Hub (...). [...] The initial conditions x(0) of the dynamical variables on vertices are set with random values from [0, 25]. We use the Runge Kutta Fehlberg (RKF45) solver to generate x(t1), x(t2), . . . , x(t80) as the training time series. [...] We devise a warm-up schedule to adapt the loss function dynamically. At the early stage of training, the model parameters are strongly affected by time-series data points closer to the initial value, which results in smaller integration errors. More specifically, we adopt a weighting schedule for the loss function with evolving weight function wk at epoch k as follows: wk(t) = e t/τk, where τk monotonically increases with the number of training epochs k.