9/8/17

I think I've finally settled on an idea, although I don't know how feasible it is. To explain this idea, I'll use an example of a game that uses machine learning: Google's QuickDraw. QuickDraw is a game in which you are given a word to draw a picture of, and the computer tries to guess what you're drawing. The computer does this using what is called a neural network, consisting of many artificial "neurons" that link together like actual neurons in the human brain. The neural network guesses what you're drawing by analyzing the drawings of previous players and seeing how closely they match what you're drawing. It's pretty amazing what neural networks are capable of. I considered applying this same neural network structure to an interesting problem. I'm considering creating a neural network that analyzes many, many recipe lists from different cookbook recipes, and attempts to create new recipe lists based off of the examples I fed it. This would allow me to analyze the neural network's recognition of flavor and ingredient combinations. Although, this does sound like quite the difficult task. I will need mentorship.