rushhan / Attention-based-GNN-reinforcement-learning-for-Knapsack-Problem

Attention-based-GNN-reinforcement-learning-for-Knapsack-Problem

Repo for knapsack problem solution based on reinforcement learning (Attention Based Graph Neural Network)

To run:

Simply execute the jupyter notebook (main_knapsack)

Notes:

I have checked proper execution for 20 nodes (items-values pairs).

TO_Do:

1. Change the names of files and function suitable for knapsack This is modified version of original vrp repo, so I have kept the changes minimal as they wont affect functionality, but might be confusing. SInly follow the jupter notebook.

Further Notes:

This is modified version of original repo .src Changes:

1. Modify VRP to Knapsack Problem
2. Data Generation suitable for knapsack problem (This could be modified by the use as needed to fit for given problem)
3. Loss for the selected item_value pair