rose-stl-lab / autostpp

Automatic Integration for Neural Spatio-Temporal Point Process models (AI-STPP) is a new paradigm for exact, efficient, non-parametric inference of point process. It is capable of learning complicated underlying intensity functions, like a damped sine wave.

Auto-STPP

✨Automatic Integration for Neural Spatiotemporal Point Process✨

| Introduction

Automatic Integration for Neural Spatio-Temporal Point Process models (Auto-STPP) is a new paradigm for exact, efficient, non-parametric inference of spatiotemporal point process.

| Citation

[2310.06179] Automatic Integration for Spatiotemporal Neural Point Processes

@article{zhou2023automatic,
  title={Automatic Integration for Spatiotemporal Neural Point Processes},
  author={Zhou, Zihao and Yu, Rose},
  journal={arXiv preprint arXiv:2310.06179},
  year={2023}
}

| Installation

Dependencies: make, conda-lock

make create_environment
conda activate autoint-stpp

| Dataset Download

python src/download_data.py

| Training and Testing

Specify the parameters in configs/autoint_stpp.yaml and then run

make run_stpp config=autoint_stpp