TIMING: Temporality-Aware Integrated Gradients for Time Series Explanation

TIMING: Temporality-Aware Integrated Gradients for Time Series Explanation
Hyeongwon Jang*, Changhun Kim*, Eunho Yang (*: equal contribution)
International Conference on Machine Learning (ICML), 2025 (Spotlight Presentation, 313/12107=2.6%)

Introduction

Official implementation for TIMING: Temporality-Aware Integrated Gradients for Time Series Explanation. TIMING is implemented in PyTorch and tested on different time series datasets, including switch-feature, state, Mimic-III, PAM, Epilespy, boiler, freezer, and wafer. Our overall experiments are based on time_interpret, ContraLSP, TimeX++, WinIT. Sincere thanks to each of the original authors!

Installation instructions

conda create -n timing python==3.10.16
conda activate timing
pip install -r requirement.txt --no-deps

The requirements.txt file is used to install the necessary packages into a virtual environment.

To test with switch-feature, additional setup is required.

git clone https://github.com/TimeSynth/TimeSynth.git
cd TimeSynth
python setup.py install
cd ..
python synthetic/switchstate/switchgenerator.py

MIMIC-III Dataset Setup

To use MIMIC-III dataset, you need to download the original dataset and set up a PostgreSQL database.

1. Download MIMIC-III Dataset

• Download from PhysioNet (requires credentialed access and CITI training)
• Sign the Data Use Agreement (DUA) and download all CSV files

2. Load into PostgreSQL

• Clone the MIMIC-III code repository:

git clone https://github.com/MIT-LCP/mimic-code.git
cd mimic-code/mimic-iii/buildmimic/postgres

• Follow the repository instructions to create database mimic with schema mimiciii and load CSV files

3. Configure Database Connection

Edit datasets/mimic3.py (around lines 151-157) to set your PostgreSQL connection:

dbname = "mimic"
schema_name = "mimiciii"
# Set host, user, and password as needed

Reproducing experiments

We have divided our experiments into two categories: Synthetic and Real.

All experiments can be executed using scripts located in scripts/real, scripts/hmm, or scripts/switchfeature.

This is an example execution for MIMIC-III (ours)

bash scripts/real/train.sh
bash scripts/real/run_mimic_our.sh
bash scripts/real/run_mimic_baseline.sh

Due to differences between our training environment and the released code, the paper’s results may not be fully reproducible with the training scripts alone. All XAI evaluations were conducted on a single NVIDIA RTX 3090 GPU (24GB VRAM). We have publicly released all model checkpoints, except those trained on the restricted-access MIMIC-III dataset.

All results will be stored in the current working directory.

And then save parsing results:

python real/parse.py --model state --data mimic3 --top_value 100
python real/parse.py --model state --data mimic3 --experiment_name baseline --top_value 100

All parsed results will be saved in the results/ directory.

Citation

@inproceedings{jang2025timing,
  title={{TIMING: Temporality-Aware Integrated Gradients for Time Series Explanation}},
  author={Jang, Hyeongwon and Kim, Changhun and Yang, Eunho},
  booktitle={International Conference on Machine Learning (ICML)},
  year={2025}
}