What are "bottlenecks" in the context of neural networks? 

This term is mentioned, for example, in [this TensorFlow article][1], which also uses the term "bottleneck values". How does one calculate bottleneck values? How do these values help image classification? 

Please explain in simple words.


  [1]: https://web.archive.org/web/20180703133602/https://www.tensorflow.org/tutorials/image_retraining