CSAIL researchers show that even “untrainable” neural nets can learn effectively when guided by another network’s built-in biases using their guidance method.