Mid-Project Deliverables:
Design Review - Reflection
Feedback and Decisions
We received good feedback regarding which directions to explore further for the remainder of out project. Everyone seemed to think that it would be more interesting to make a neural network which directly outputs moves, rather than outputting board evaluations. This method will completely eliminate the min-max structure of our original algorithm, and will instead do all of its decision making through a neural network. The other option we presented was to continue using the min-max structure, but to replace the algorithmic board evaluation at the base of the evaluation tree with an evaluation function which would be developed using a neural network and a database of past chess games. Based on the feedback from the code review, we have decided to pursue the first method, which will use neural networks for the entirety of the decision making.
Review Process Reflection
Our review accomplished what we wanted to do. In this review we gave a more appropriate amount of background information so that our audience understood what we had told them and still had enough information to give us useful feedback. We also stuck to our agenda better this time because the review went the way we thought it would and we were able to get more audience participation. Because of where we were in our project, we didn't have a whole lot that we wanted to get from the code review, but we did get answers to the questions we presented, and the answers we got will definitely help us move forward with our project