Choosing tensor size
I'm new to deep learning, and currently working on a deep-learning program to apply on bubble shooter game. I'd be happy if someone can give me a lead, what's the most efficient way to choose the states when approaching a problem? for example, I considered checking the color of the ball in the cannon, and then scanning the environment and looking where there's the same color. consequently, adjusting the canon in the necessary direction. I still have a bit trouble transferring it from pseudo code to real code, but before I do that, I wanted to consult whether it's a good idea or should I consider another approach.