Architecture Module (radionets.core)

Introduction

radionets.architecture contains architecture classes and functions for the deep-learning framework.

Submodules

• Activation Functions (radionets.architecture.activation)
• Architectures (radionets.architecture.archs)
• Blocks (radionets.architecture.blocks)
• Layers (radionets.architecture.layers)
• Uncertainty Architectures (radionets.architecture.unc_archs)

Reference/API

radionets.architecture Package

Classes

BottleneckResBlock(in_channels, out_channels)	Three-convolution layer deep residual neural network building block.
Decoder(in_channels, out_channels[, ...])	Decoder block for UNets.
Encoder(in_channels, out_channels[, ...])	Encoder block for UNets.
GeneralELU([add, maxv])
GeneralReLU([leak, sub, maxv])
Lambda(func)
LocallyConnected2d(in_channels, ...[, bias])	A 2D locally connected layer implementation.
NNBlock(in_channels, out_channels[, ...])	Base block for neural networks.
SRBlock(in_channels, out_channels[, ...])	Default SRResNet building block.
SRResNet()
SRResNet18()
SRResNet18Complex()
SRResNet18AmpPhase()
SRResNet34()
SRResNet34AmpPhase()
SRResNet34_unc()
SRResNet34_unc_no_grad()
Uncertainty(img_size)
UncertaintyWrapper(img_size)