This pretty much my same pick. I'm completely blown away by the possibilities of a combination of gradient descent and clever loss/architectures. After learning

This pretty much my same pick. I'm completely blown away by the possibilities of a combination of gradient descent and clever loss/architectures. After learning how output shape and loss definition can result in wildly different tasks such as landmark detection, classification, or single shot multibox detection(bounding boxes AND classification wuuut?!) it really gets the imagination running wild.