What's in the black box guys?🔥
Good day to all! 👋
I know that the topic is not new, but it is definitely relevant for everyone who works in the field of machine learning development.
The fact is I am so "concerned" about Black Box in ML that wrote a review dedicated to this isssue;)
I highlighted such areas in it:
✔️ In What Cases Do Black Box Machine Learning Models Work Best?
✔️ What Problems are Connected With the Black Box?
✔️ How to Resolve them?
💥 Here it is: https://y-sbm.com/blog/black-box-in-machine-leraning/ 👈
I would be grateful for the support to improve this article, as well as ready for any kind of constructive criticism.
A good example, as you have noticed, is the ATMSeer program. It will greatly increase the level of trust in such systems.
But one cannot deny the fact of developing a more transparent system for the user. Taken together, this will give a much better effect.
For me, developing the Black Box principle is really complicated, but usually justifies the time spent.
I completely agree, but before that it is necessary to plan everything well, for accuracy of work.
Truth. This is a plus that the user doesn’t have access to the internal work of the code. This is important enough to protect such systems.
Yes, this is one of the important aspects. It is also useful for protecting certain technical decisions that have been made during development from third parties.
In fact, the problem of distrust in AI is quite significant. People usually do not trust things that they do not know how they work, and this is one such case.
Of course, there is always a risk, but this concept is still used.
Usually this is not a nice factor, but there are ways to deal with it. By the way, I have already discussed this in the comments above
This comment was deleted 7 years ago