standard_networks.h File Reference

Declares ready-made NeuralNetwork architectures for common learning tasks (regression, classification, forecasting, image classification, auto-association, ResNet, VGG-16, text classification, Transformer). More...

#include "neural_network.h"

Go to the source code of this file.

Classes
class	opennn::ApproximationNetwork
	Standard regression (function approximation) MLP. More...
class	opennn::ClassificationNetwork
	Standard tabular classification MLP. More...
class	opennn::ForecastingNetwork
	Standard time-series forecasting MLP. More...
class	opennn::AutoAssociationNetwork
	Standard auto-encoder for outlier and novelty detection. More...
class	opennn::ImageClassificationNetwork
	Standard convolutional image classifier. More...
class	opennn::SimpleResNet
	Compact residual network for image classification. More...
class	opennn::VGG16
	VGG-16 architecture (Simonyan & Zisserman, 2014) for image classification. More...
class	opennn::TextClassificationNetwork
	Standard text classification model. More...
class	opennn::Transformer
	Encoder-decoder Transformer (Vaswani et al., 2017) for sequence-to-sequence modeling. More...

Namespaces
namespace	opennn

Detailed Description

Declares ready-made NeuralNetwork architectures for common learning tasks (regression, classification, forecasting, image classification, auto-association, ResNet, VGG-16, text classification, Transformer).

Each class is a thin NeuralNetwork subclass whose constructor wires the appropriate scaling, hidden, output and unscaling layers given a few shape arguments. They serve as starting points; users can still mutate the returned network with add_layer() / connect_layers() before compile().