omonimus1/rocket_landing

🚀 CNN Parameter Tuning for Rocket Landing system - Coursework of Computation Intelligence at Napier University

Rocket landing

The current repository contains the SET10107 Computational intelligence coursework, given by Edinburgh Napier University. Having a CNN used for the landing system of the a rocket, the main goals are:

• Prune training parameters to optimize the weights of a Multi-layer Multi-layer Perceptron, applied to the landing system of a rocket
• Get lowest fitness score, produced by the average of multiple runs of the same Evolutionary algorithm;

Technologies

Evaluation

When running StartGui.java, the running of the UI will slow down the execution of the algorithm, reporting the trends of fitness score just until the 10.000th running; Instead, this project runs the EA 20.000 times; To optimize the training time, and visualize final results, you may want to start the training by running StartNoGui.java

Results obtained:

• Average fitness score: 0.145 (against 14.8 produceded by the given project template).

🔗 Links - Get my Linktree

Apply for a mentorship session with me (get a 7 days free trial)