And while I was driving I had another idea about applying reinforcement learning to some small and fun project. I called it mi-mi-mi meter. mi-mi-mi means cute, so this system will run a random search by pictures on the internet and user will rate some of them that he considers cute. That's how the user teaches the system. When the system is trained for some time it will recommend the best cute pictures from the internet without corrections from the user.
Published: