ComplexValuedCNN

Complex CNN in Pytorch, implementation of the paper Deep Complex Networks : https://arxiv.org/abs/1705.09792

Modules implemeted :

• Complex Conv2d
• Complex ConvTranspose2d
• Complex AvgPool2d
• Complex Linear

Activation function :

• zRelu
• CRelu

Usage :

1. Define your Complex input using the class CPLX. x = CPLX(real_part, imaginary_part)
2. Build your model using src modules.
3. Use the loss function in utils.py to train your model. if the labels are real valued, set use_magnitude = True

ToDo

• Complex UNet