Conv1d_Modules_tf

This repository contains 1-dimmensional cnn modules for regression task. I used these modules for time-series prediction such as blood pressure(BP) prediction using electrocardiogram(ECG) and ballistocardiogram(BCG). I tried to follow the original paper's implementation [1][2][3] as closely as possible, but I made some improvement in certain parts. [4]

Supported Modules

1. VGG_Net_1D [1]
2. ResNet_1D [2]
3. RepVGG_Net_1D [3]

How To Use :

git clone https://github.com/NoahYn/Conv1d_Modules.git
cd Conv1d_Modules
python [file name to run]

you can choose sub_module in each file.

Specification :

1. VGG_Net

2. ResNet

3. RepVGG

TODO :

---

Implement DenseNet_1D, ResNeXt_1D, EfficientNet_1D ...

Add explanation

Add example to test

Add pytorch version

References

---

[1] Simonyan, K., & Zisserman, A. (2015). Very Deep Convolutional Networks for Large-Scale Image Recognition. https://arxiv.org/abs/1409.1556.

[2] He, K., & Zhang, X., & Ren, S., & Sun, J. (2015). Deep Residual Learning for Image Recognition. https://arxiv.org/abs/1512.03385

[3] Ding, X., & Zhang, X., & Ma, N., & Han, J., & Ding, G., & Sun, J. (2021) RepVGG: Making VGG-style ConvNets Great Again https://arxiv.org/pdf/2101.03697

[4] Ioffe, S., & Szegedy, C. (2021). Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. arXiv.org. Retrieved 30 August 2021, from https://arxiv.org/abs/1502.03167.

Related Works

https://github.com/Sakib1263/VGG-1D-2D-Tensorflow-Keras https://github.com/DingXiaoH/RepVGG/tree/main https://github.com/hoangthang1607/RepVGG-Tensorflow-2