Play 9 Man's Morris against the QUANTUM AGENT!
We researched possibilities to get an computer driven opponent for 9 Man's Morris.
Machine learning including quantum computing layers seems viable but needs better training accuracy.
Currently you are able to play against a Quantum Agent, getting the moves from a real quantum computer. We modeled an Ising-Model for the game, which the Quantum-Agent tries to solve in order to calculate the best move.
- Model the problem for a quantum computer and let it solve it;
- Trying Hybrid Quantum/Classical Neural Network.
- Understanding the work from last year and link it all together. We had to salvage quite a lot and re-implement into our own code. However we were able to improve it;
- Understanding how to model problems for Quantum Computing;
- Finding Morris?!
Phase 1 and Phase 2 is now fully playable, even if the opponent does not always make the smartest moves.
We learned a lot about how Quantum Computing works and how to model problems for quantum computers.
Get the Quantum Agent working for Phase 2 and 3.
Improve the Quantum Agent for Phase 1.
Try to improve the results of the neural network using a quantum layer. The current accuracy is not as good as with a common approach. But we are sure it can be improved.