Answer the question
In order to leave comments, you need to log in
Is it possible to create a neural network that predicts numbers from a pseudo-random sequence?
Let's say we give the first hundred numbers as input, and 101 numbers as output. And yes. How do random number generators work in personal computers?
Answer the question
In order to leave comments, you need to log in
For a good PRNG, no, because in a good PRNG, there is no visible pattern in the output data, if something can be found there by a neural network, then it is bad by definition. But if your PRNG gives out a repeating sequence of 10 elements, then yes.
For a bad PRNG, you can use a neural network to detect strange attractors, for example. An example of using strange attractors for PRNG analysis is in this article:
lcamtuf.coredump.cx/oldtcp/tcpseq.html
PS and do not confuse random and pseudo-random number generators. Random number generators usually use some kind of physical principles, such as noisy diode generators. PRNG is usually algorithmic + entropy collection can be used.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question