Skip to main content

Project Synopsis

Genetic Algorithm for Deep Learning

This project uses genetic algorithms to optimize Deep learning parameters. 


Problem being addressed

Deep Learning and machine learning models require vast amounts of unique hyperparameters to develop the most optimized model for a solution. Finding these hyperparameters requires immense trial and error, which leads to lost time that could be spent elsewhere in the software development or research pipeline. Also, many of the hyperparameters in these models can have a high barrier for understanding to one who doesn’t have strong knowledge in statistics or machine learning.



The proposed solution is to use a type of algorithm that works very well on optimization problems. This algorithm is known as genetic algorithms, which takes its roots from nature. The team is working on constructing a genetic algorithm that can intake a deep learning or machine learning model and its desired hyperparameters as parameters to the genetic algorithm and then return the most optimal hyperparameters for the model.


Tools and Technologies

Language: Python

Libraries: Tensorflow, Keras, Scikit-Learn, Numpy, Matplotlib

Package Management: Anaconda

Version Control: Gitlab

IDEs: Pycharm, Visual Studio