What is the dying ReLU problem in neural networks?

Related questions

1. Why is CNN deep learning?
2. Why does CNN use image classification?
3. Which models are best suited for recursive?
4. Why is neural network nonlinear?
5. Why is CNN used?
6. What is the method to overcome the decay of information through time in RNN known as?
7. How do recurrent neural networks work?
8. Is deep learning better than machine learning?
9. Why is Lstm better than RNN?
10. Is sigmoid better than ReLu?

Tags

1. #longer
2. #during
3. #descent
4. #problem
5. #training
6. #return
7. #networks
8. #units
9. #likely
10. #stochastic
11. #outputs
12. #refers
13. #neural
14. #useful
15. #rectified
16. #linear
17. #gradient
18. #dying
19. #given
20. #input