This seventy-page booklet (suitable for a day's read) aims to give the reader a quick understanding regarding the topic of safety for machine learning (ML), where among many techniques in machine learning, the focus is on deep neural networks (DNN). The intended readers are primarily university students but can also include safety engineers, machine learning engineers, or researchers from other fields who wish to develop safe ML applications.

Neural network safety is, by definition, an intersection between machine learning and safety engineering. Consequently, it is uncommon to have people holding expertise in both fields. It is my belief that bringing AI/ML from running demos to safe products requires people from different sides to "meet in the middle". This booklet may introduce people on one side to know the language on the other (e.g., ML people to know safety) and imagine how knowledge from both sides can be combined.