There exist various blog posts describing the time complexity of Fully Connected Neural Networks (1, 2, 3, 4); Convolutional Neural Networks (CNN) (5) and of Long Short-Term Memory (LSTM) networks (6). However, this question seems to be absent in the academic literature or at least I couldn't find it. Furthermore, papers that go towards this question seem to consider mostly experimental results (7, 8).

Therefore, my question is:

  • Is there a literature on the time complexity of Neural Networks that gives analytic expressions in the fashion of $\mathcal O(\cdot)$?
  • If this is not the case this is because...
    • ...(1-6) is common sense?
    • ...it is not that easy to answer and the blog posts only present a simplified form of the problem?
    • ...something different?

Although I consider the posts mentioned above very useful, there are various reasons why one would prefer to cite a paper rather than a website (10)!

  • $\begingroup$ We have other posts other than the only one from our AISE site that you're linking to. See this, this, this, this, etc., some of them answered by myself. So, I recommend that you go through all these posts and update yours with links to them, and revise your question, if necessary. $\endgroup$
    – nbro
    Commented Jul 6, 2022 at 17:23


You must log in to answer this question.