evolutionary-algorithms-from-scratch

Designing variations of several evolutionary algorithms from scratch without using external ML-libraries.
(current version: 1.0)

Author: Siddharth Yadav (syntax-surgeon)

evolutionary-algorithms-from-scratch-v1.ipynb

• Contains version-1.0 of the main program which tries to implement an evolutionary algorithm to optimize the classic knapsack problem
• The evolutionary algorithm is compared to a brute force approach to highlight the performance gain
• Several statistics of the search are displayed in real-time such as iterations, generations, success rate etc.
• Automated graph generation is available in two options: value increase per generation and one-hot encoding scheme
• A testing system is also employed to check the overall success rate over multiple runs

Value (fitness) increase per generation graph

One-hot encoding scheme plot

Testing phase results