After weeks/months of construction noise, I haven’t really gotten to do any kind of coding. I felt rusty, so I decided to look up neural networks and create a simple one. The reading is a bit more complex than this thing actually is, and once the code was in place some interesting results happened.

My task: Let a create (a red square) hunt for food (green squares)

Note: No Backwards propagation means it doesn’t get smarter. It just connects synapses with random values. Inputs are directions to the target food, as well as the distance. Just refresh if it starts running away from the food.

See it in action here (CLICK)