Deep learning is getting importance in almost every field. Read number of articles but still confused about its basics. Can anyone help to go more simple but in depth analysis?
As per my knowledge in neural networks in Image procesding applications. . In practice, when solving big problem by making into sub-problems, we usually start by figuring out how to solve sub-problems, and then gradually integrate the solutions. In other words, we build up to a solution through multiple layers of abstraction . The layers of abstraction seem likely to give deep networks a compelling advantage in learning to solve complex application easier.
In simple words " a big task is divided into sub-tasks, such that each sub-task is handled by each layer in multiple layer NM finally giving the final result in a easy mode.
One thing to realize here is that neural network systems is an approximation or a sort of black box, which has learnt the characteristics of the system. So whatever inputs you will give the NN, the outputs will be based on the type of learning. One way to improve the learning is to keep it adaptive so that system continues to learn about the system with every input and output that is generated. My early research was in this area but after much research work, I have come to the conclusion that no matter how good the learning is, it will still be an approximation so controls or any other strategy cannot solely depend upon it. People have started to move on from these heuristic techniques.