The multi-objective KnapSack is a trending combinatorial optimisation problem that can be solved with metaheuristics, but this is computationally difficult and costly. Several studies have shown that Machine Learning can be a good alternative to solve such problems by predicting the optimal solution and this work is an initiative to materialize this theory.
Before execution, place the provided folder 'mkpInstances' under users/maily or users/'username'
It is importnant to insist on the fact this is an initiative to leverage artificial intelligence (Deep Learning) in the field of combinatorial optimization, we are open for any suggestions or critics that would help improve it.
This work was done in an academic frame and under the supervision of Mrs. Imen Ben Mansour (https://scholar.google.fr/citations?user=E8gXmGcAAAAJ&hl=fr) in Esprit University of Tunisia.
Many files, code blocks and ideas were taken from open sources. Links are available in the notebook and in the project report.