Hi everyone, this is an old project that is finally becoming public. Back when I was deep into experimenting with neural networks I created this game and put an neural network to play it (so it could learn how to shoot a target). The mechanics are very basic, no air resistance, the enemy is always on the same spot but, the neural network is fresh each game, it only learns while playing against you. On desktops arrows left and right rotate the cannon, up and down increase and decrease the strength of the bullet. Spacebar shoots.

On mobiles (better on landscape orientation) dragging your finger up or down controls both things, but if you're dragging of the left side of the screen it controls the strength, and on the right side, the angle. Taping on the screen shoots.

By the way, you are the gray one.

Let me know what you think.