refine

microsoft · Dec 6, 2023 · 9229a84 · 9229a84
1 parent 7837855
commit 9229a84
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 25 deletions.
diff --git a/docs/configs/contiformer_classfication.yml → docs/configs/contiformer_classification.yml b/docs/configs/contiformer_classfication.yml → docs/configs/contiformer_classification.yml
diff --git a/docs/contiformer.md b/docs/contiformer.md
@@ -1,33 +1,29 @@
-# CinC2020
+# Contiformer
 
 This document gives an example on how to use the PhysioPro framework for [ContiFormer: Continuous-Time Transformer for Irregular Time Series Modeling](https://seqml.github.io/contiformer/).
 
-The example is based the solution of the 1st ranked team [Prna](https://ieeexplore.ieee.org/document/9344053).
+## Classification
 
+Here we take the `Heartbeat` classification task from [UEA & UCR Time Series Classification Repository](http://timeseriesclassification.com/dataset.php) as an example.
 
-## Scripts
-
+1. Download the dataset
 ```bash
-# prepare data and features
-cd data/
-
-# training data
-mkdir -p CinC2020/training_data
-wget -r -N -c -np https://physionet.org/files/challenge-2020/1.0.2/ # 10G
-find physionet.org/files/challenge-2020/1.0.2/training -name "*.hea" -exec mv {} CinC2020/training_data \;
-find physionet.org/files/challenge-2020/1.0.2/training -name "*.mat" -exec mv {} CinC2020/training_data \;
+cd PhysioPro
+mkdir data
+wget http://www.timeseriesclassification.com/aeon-toolkit/Archives/Multivariate2018_ts.zip -P data
 
-# features
-wget https://physionet.org/static/published-projects/challenge-2020/1.0.2/sources/Prna.zip
-unzip Prna.zip && rm Prna.zip
-mv Prna/physionet2020-submission/feats Prna/physionet2020-submission/records_stratified_10_folds_v2.csv Prna/physionet2020-submission/top_feats.npy CinC2020
-rm -rf Prna
+unzip data/Multivariate2018_ts.zip -d data/
+rm data/Multivariate2018_ts.zip
+```
 
-# weights
-wget https://raw.githubusercontent.com/physionetchallenges/evaluation-2020/master/weights.csv -P CinC2020
-
-# run training
-cd ..
-mkdir -p outputs/cinc2020
-python -m physiopro.entry.train docs/configs/cinc2020.yml
-```
+2. Run `Heartbeat` classification task with `TSRNN` model
+```bash
+# create the output directory
+mkdir -p outputs/Multivariate_ts/Heartbeat
+# run the train task
+python -m physiopro.entry.train docs/configs/contiformer_classification.yml
+# tensorboard
+tensorboard --logdir outputs/
+```
+
+The results will be saved to `outputs/Multivariate2018_ts/Heartbeat` directory.
diff --git a/setup.py b/setup.py
@@ -68,6 +68,7 @@ def setup():
         install_requires=[
             "torch==2.0.1",
             "torchvision==0.15.2",
+            "torchcde==0.2.5",
             "utilsd==0.0.18.post0",
             "pandas>=2.0.3",
             "scipy>=1.10.1",