Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
PyTorch implementations of several SOTA backbone deep neural networks (such as ResNet, ResNeXt, RegNet) on one-dimensional (1D) signal/time-series data.
| Date | Stars |
|---|---|
| 2026-07-31 | 546 |
| 2026-08-01 | 547 |
| 2026-08-02 | 547 |
| 2026-08-06 | 547 |
| 2026-08-08 | 546 |
| 2026-08-11 | 546 |
| 2026-08-12 | 546 |
| 2026-08-18 | 546 |
| 2026-08-20 | 547 |
| 2026-09-07 | 548 |
| 2026-09-20 | 548 |
Today
— stars today
This week
— stars this week
This month
+1 stars this month
Momentum
0.0
growth rate 0.00%/day
# Introduction
PyTorch implementations of several SOTA backbone deep neural networks (such as ResNet [1], ResNeXt [2], RegNet [3]) on one-dimensional (1D) signal/time-series data.
If you use this code in your work, please cite our paper
```
@inproceedings{hong2020holmes,
title={HOLMES: Health OnLine Model Ensemble Serving for Deep Learning Models in Intensive Care Units},
author={Hong, Shenda and Xu, Yanbo and Khare, Alind and Priambada, Satria and Maher, Kevin and Aljiffry, Alaa and Sun, Jimeng and Tumanov, Alexey},
booktitle={Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining},
pages={1614--1624},
year={2020}
}
```
# Usage
```
# test on synthetic data, no data download required
python test_synthetic.py
# test on PhysioNet/CinC Challenge 2017 data
# need prepare data first, or you can download my preprocessed dataset challenge2017.pkl from https://drive.google.com/drive/folders/1AuPxvGoyUbKcVaFmeyt3xsqj6ucWZezf
# Please see comment in code for details
python test_physionet.py
# training or serving model with Ray [4], need install Ray first: https://github.com/ray-project/ray
# Please see comment in code for details
python test_ray.py
```
model_detail/ shows model architectures
In ```trained_model```, we also provide a model ```model.pth``` trained using Challenge 2017 data. The model's parameters is as follows:
```
model = Net1D(
in_channels=1,
base_filters=64,
ratio=1.0,
filter_list = [64, 160, 160, 400, 400, 1024, 1024],
m_blocks_list = [2, 2, 2, 3, 3, 4, 4],
kernel_size=16,
stride=2,
groups_width=16,
verbose=False,
n_classes=4)
```
We can directly load this model by:
```
model = torch.load('model.pth')
```
The below image shows the change of training loss with epoch.
Confusion Matrix on Validation Set:
| | Normal | AF | Others | Noisy |
| ---- | ---- | ---- | ---- | ---- |
|**Normal** | 0.878| 0.017 | 0.069 | 0.036 |
| **AF** | 0.032 | 0.861 | 0.077 | 0.030 |
| **Others** | 0.310 | 0.126 | 0.515 | 0.049 |
| **Noisy** | 0.107 | 0.058 | 0.020 | 0.815|
Other Metrics on Validation Set:
|AUC|F1-score|Accuracy|
|----|----|----|
|0.931|0.762|0.769|
# Requirements
Required: Python 3.7.5, PyTorch 1.2.0, torchsummary
Optional: Ray 0.8.0
# Application 1: ECG Classification (PhysioNet/CinC Challenge 2017)
The initial code has been used in our previous work [5,6,7] for deep feature extraction, which won one of the First place (F1=0.83) of this Challenge. The original tensorflow (tflearn) version can be found at https://github.com/hsd1503/ENCASE .
Dataset: "AF Classification from a short single lead ECG recording". Data can be found at https://archive.physionet.org/challenge/2017/#challenge-data Please use Revised labels (v3) at https://archive.physionet.org/challenge/2017/REFERENCE-v3.csv , or you can download my preprocessed dataset challenge2017.pkl from https://drive.google.com/drive/folders/1AuPxvGoyUbKcVaFmeyt3xsqj6ucWZezf .
This repository also contains data preprocessing code, please see util.py for details.
# Application 2: Health Monitoring in Intensive Care Units (KDD 20)
We built a set of models (called model zoo) for ensemble serving in Intensive Care Units. Please see more at [8].
# References
[1] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. CVPR 2016 [paper](https://arxiv.org/abs/1512.03385)
[2] Saining Xie, Ross B. Girshick, Piotr Dollár, Zhuowen Tu, Kaiming He. Aggregated Residual Transformations for Deep Neural Networks. CVPR 2017 [paper](https://arxiv.org/abs/1611.05431)
[3] Ilija Radosavovic, Raj Prateek Kosaraju, Ross Girshick, Kaiming He, Piotr Dollár. Designing network design spaces. CVPR 2020 [paper](https://arxiv.org/pdf/2003.13678.pdf)
[4] Philipp Moritz, Robert Nishihara, StephanExcerpt of 5,420 characters
Read on GitHub
46
6
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:c754bbb9b16a6e9b, llm:Repository description and README: 'PyTorch implementations of several SOTA backbone deep neural networks (such as ResNet, ResNeXt, RegNet) on one-dimensional (1D) signal/time-series data.' Topics: deep-neural-networks, ecg-classification. Provides models, training/testing scripts and pretrained model.pth for ECG/PhysioNet data.
matched fp:c754bbb9b16a6e9b, llm:Repository description and README: 'PyTorch implementations of several SOTA backbone deep neural networks (such as ResNet, ResNeXt, RegNet) on one-dimensional (1D) signal/time-series data.' Topics: deep-neural-networks, ecg-classification. Provides models, training/testing scripts and pretrained model.pth for ECG/PhysioNet data.