Data compression from lack of patterns

83 Views Asked by At

From my understanding, lossless data compression takes advantages of statistical patterns. I want to know if it is possible to compress data based on the lack of patterns, such as a random string of digits (like the kind you would find in the digits of $\pi$). The saved space comes from the complete lack of any pattern, which itself is unique. If it has already been done, I would like to know if there are any notable implementations.

1

There are 1 best solutions below

2
On BEST ANSWER

Intuitively, this might sound plausible, but, in fact, there are sooooo many more ways of being unpatterned than of being patterned that there's no way to abbreviate it losslessly.

But, if we pursue the aspect that much of the "patternless" aspect of photos and such is irrelevant to human visual perception (for example), then we can find lossy things like JPEG that "lose lots of information", but, hopefully, the huge amount of "information" that is lost is irrelevant. Just noise.

So, short summary, if all the details of the chaotic info matter, it's a problem, because you probably can't compress it. But if, for your purposes, the "random" parts are also the "irrelevant" parts, then various lossy schemes (JPEG being historically one of the first, I think) can be very good.