The future of deep learning is hard to predict at the moment; things are changing rapidly. However, I believe that if you invest your time in the present advanced topics in deep learning, you might see these areas developing prosperously in the near future.
The following sub-sections discuss some of these advanced topics that have the potential of flourishing and being disruptive in our area.
Deep reinforcement learning
Deep reinforcement learning (DRL) is an area that has gained a lot of attention recently given that deep convolutional networks, and other types of deep networks, have offered solutions to problems that were difficult to solve in the past. Many of the uses of DRL are in areas where we do not have the luxury of having data on all possible conceivable cases, such as space exploration, playing video games, or self-driving cars.
Let's expand on the latter example. If we were using traditional supervised learning to make a...
