Evol-AutoNN: Evolutionary Automation of Neural Network

Abstract

As the domain of data science keeps on growing it leads to the need for a tool that breaks the barriers for non-experts to get into the field of data science. In this paper, we introduce the project of automating the neural network with the Evolutionary algorithm in the name of Evol-AutoNN. Evol-AutoNN is a biomimicry approach to the prediction of tabular data using the well-known two biological concepts that came from nature. Biological Evaluation and the Neural Network that came from the human brain are the two concepts applied in this project. Biomimicry is a new procedure to do innovation that tries to give a strong solution to a human problem by mimicking the patterns identified in nature. It is developed to overcome the problem of finding the best initial setting for the neural network for a given problem by the user. Evol-AutoNN imitates the process followed in both natural selection and genetics. As such, this work entitles to be an important step in the automation of neural networks for tabular data.