OpenNN is a C++20 library for building, training and running neural networks, with CPU and optional CUDA execution. This is the API reference for OpenNN 26.0. For tutorials and guides, visit opennn.net.

The library is developed by Artelnics. The source code is at github.com/Artelnics/opennn.

Core classes

Class	Role
TabularDataset, ImageDataset, TextDataset, YoloDataset	Data loading, sample roles, scaling and statistics
Network	Architecture and layers; ClassificationNetwork, ApproximationNetwork and ForecastingNetwork build common models
Training	Pairs a network and a dataset with a loss and an optimizer, and trains
Evaluation	Testing on held-out data: errors, confusion, ROC and other metrics
ModelSelection, InputSelection	Architecture and input search
ResponseOptimization	Finds inputs that optimize the outputs under constraints
ModelExpression	Exports a trained model as source code

Layer types

Dense, Convolutional, Pooling, Pooling3d, LSTM, Recurrent, Embedding, MultiHeadAttention, Normalization3d, Scaling, Unscaling, Clamping, Addition, Flatten.

Optimizers

Adam, SGD, QuasiNewton (BFGS) and LevenbergMarquardt.

First program

This excerpt from the Iris example (examples/iris_plant/main.cpp) trains a classifier and evaluates it. The full example also exports the model:

#include "opennn/core/configuration.h"
#include "opennn/dataset/tabular_dataset.h"
#include "opennn/models/models.h"
#include "opennn/evaluation/evaluation.h"
#include "opennn/training/training.h"
 
using namespace opennn;
 
int main()
{
    Configuration::instance().set(Device::CPU, Type::FP32);
 
    TabularDataset dataset("../data/iris_plant/iris_plant_original.csv", ";", true, false);
 
    ClassificationNetwork network(dataset.get_input_shape(), {16}, dataset.get_target_shape());
 
    Training training(&network, &dataset);
    training.train();
 
    Evaluation evaluation(&network, &dataset);
    evaluation.print_multiple_classification_tests();
 
    return 0;
}

Navigating the reference

Class List: every class in alphabetical order.
Class Index: alphabetical index by name.
File List: by header file.

Resources

Tutorials: https://www.opennn.net/tutorials/
Building OpenNN: https://www.opennn.net/tutorials/building-opennn/
Licensing options and commercial enquiries: https://www.opennn.net/licensing/
License: GNU Affero General Public License, version 3 only, with the additional permissions in LICENSING.md, or a commercial license from Artelnics. OpenNN 8.x and earlier remain under the GNU LGPL 2.1 or later.
Issues and contributions: https://github.com/Artelnics/opennn