Is it feasible to compress a real-time updated neural network and deploy it on terminal devices?
Is it feasible to compress a real-time updated neural network and deploy it on terminal devices?