SeqLink: A Robust Neural-ODE Architecture for Modelling Partially Observed Time Series

Note: We are still working on organizing this code. Some features and documentation may be incomplete or subject to change.

This repository contains the code for our enhanced Neural-ODE Architecture. The code is written in PyTorch and follows the methodology described in our paper available on OpenReview\SeqLink.

Model Overview

We build our code on the publicly available code for ODE-RNN at Yulia Rubanova's GitHub For the baselines (RNN-VAE, Latent ODE, and ODE-RNN) we follow the implementation available at Yulia Rubanova's GitHub. For the CDE model, we follow the implementation available at Patrick Kidger's GitHub. For TSMixer we follow the implementation available at ditschuk's GitHub.

Datasets

All datasets used are in the Dataset folder. Including original data and the .pt format to be used for the ODE-RNN model

The learn representation generated using ODE-RNN is saved in DataRep/ fplder. To regenerate the we recommend you follwoing the instructions of the original code repository

Tha attention and payramid module codes are in scr.

to generate the final prediction Run SeqLink.py

Repository Structure

• src/: Contains the source code for the model, attention mechanism, and pyramid sorting.
• data/: Example data files used for training and testing.
• notebooks/: Jupyter notebooks demonstrate the use of the model and visualize results.
• docs/: Documentation and additional resources.

Installation

To install the required dependencies, run:

pip install -r requirements.txt