For example take this paper[PDF]. If I wanted to add Weather degradation into a clear photograph would it be worth it to look at that paper and try to reverse it? Also is there any specific approach to reversing an algorithm if it is at all possible?
views:
163answers:
4
A:
Although that particular technique (of reversing an algorithm) can be extremely useful, I would think instead that finding the correct technique you were looking for would be more beneficial. As far as reversing an algorithm, think of how you reverse an equation in Mathematics, and do something similar - i.e. at each step in the algorithm, attempt to do the reverse step. Note that in some algorithms (lossy compression comes to mind) this is not always possible, as some of the data is lost because of the algorithm.
aperkins
2009-07-24 17:09:39
A:
Yes. Just because something disagrees with your philosophy of the world does not make it a bad thing to read. It in fact makes it a good thing to read. Only reading texts which subscribe to your belief system is the equivalent of a person in power surrounding themselves with a group of "yes" men.
You should be constantly challenging your strongest beliefs to ensure you are on the right path.
JaredPar
2009-07-24 17:10:11
Its not that I disagree, its opposite in its end goal. It removes weather degradation from photographs, I want to add it.
Soldier.moth
2009-07-24 17:12:07
+7
A:
There is no general approach. Take multiplying primes vs factoring into primes. It's dead easy to take:
A * B = C
and calculate C from A and B. Getting A and B from C is not so easy, even though there's only one solution.
You can see why trying to reverse (in general) the effects of any algorithm isn't so simple!
Yes, in certain cases it may be one useful technique, but if looking at the paper doesn't immediately give you an idea for how you might reverse it, I'd take it as a sign you should probably look for what you're actually after, first.
Draemon
2009-07-24 17:13:40
"even though there's only one solution.", there actually might be more then one solution if the problem isn't linear.
nlucaroni
2009-07-24 17:20:20
Not quite sure what you mean by linear in this context. There's only one prime factorization of any number.
Draemon
2009-07-24 17:32:18
didn't see the context of primes, BUT what makes you think this is something like factoring primes? a projection for, example, of R3 -> R2 has more then one solution in reverse.
nlucaroni
2009-07-24 19:19:11
... the zeros of a function, have multiple solutions, and the reverse (knowing the zeros and obtaining the function), also has infinite solutions. So, the problem is even harder then what you describe.
nlucaroni
2009-07-24 19:28:20
My point was that factoring primes is reversing a really really simple operation (multiplying them), and it's incredibly hard... so assuming that he's *not* starting with a problem this simple, he definitely can't assume the reverse is going to be easy or even possible. As you say, the reality is even worse.
Draemon
2009-07-24 20:35:52
+1
A:
Draemon is correct, there is no general approach. In your case, I would read the paper.
Whether you want to add or remove weather degradation, you need to know what weather degradation looks like. That paper will have information on that, and might have an easy answer.
An example might be a paper on breaking a substitution cipher. You can certainly read the paper to learn how to encrypt something with a substitution cipher.
Michael Sofaer
2009-07-24 22:37:48