Use CNN (preferably in Tensorflow) to detect Hamiltonian Circuit solution. Input layer: a graph, Target output layer: Yes/No (HC). You will need to create thousands of training graphs with known labels Yes/No. At least 500 yes, 500 no, for training. At least 50 yes, 50 no, for testing.
end result. A report discussing (1) your graph generation (or, selection, if you generate randomly and use your previous algorithm to select yes/no), (2) deep learning network and parameters, (3) results (accuracy graph against epoch, for both training and test sets), and (4) CNN code (Tensorflow). Jupyter IDE is typically used by most data scientists.
due in 3 days, 70 hours