The Convolutional Neural Network Hyper-Parameterisation Genetic Algorithm Optimisation Framework (CNNHGAO) is a system that autonomously creates and fine-tunes deep learning model architectures through hyper parameterisation optimisation and genetic algorithms optimisation techniques.

Genetic algorithms were implemented in order to allow the system to choose effective hyperparameters (such as layers, nodes, etc.) for the architecture. The system generates the neural network architecture through a process of injecting neural network configurations that have been determined through evolutionary computing using genetic algorithms. The injected configurations utilise code generation within the framework.

In addition to this, the CNNHGAO framework will enable the prediction of the architecture of deep learning models through the use of machine learning strategies.

The CNNHGAO framework will allow companies that use computer vision to bypass the notoriously difficult process of manually configuring these models in the creation of products such as self-driving cars or facial recognition systems.

Currently, there is considerable active research in this area (meta-learning) by a number of tech giants such as Google, Microsoft, etc.