Re: Effective Lossy Compression of Bitstream


Thanks for the posts.

IOW, are you sure you can *always* reconstruct the missing data? If so,
we are in the lossless data compression domain of a redundant source,
which is a different business. (Note that if your source has no
rendundancy built-in, then there is no way of reconstructing this
missing piece of information and you always have loss. Can you tolerate
this?)

The last and most final important question to know is, "What are you
compressing and what do you consider NOISE?"

As I stated previously, I'm coding a program behavior. When the
program processes it's data (which is a calculation), it can "change"
it's state. So, when the program encounter an "1", it change its way
of processing the data. When I encounter the zero, it continues
processing the same way as before.

I told the program can recover the information because when some
things happen while processing the data, the program can identificate
that it has to change it's state, but it'll be one step AFTER it
should, so the processed data get's processed not the way it should
(the calculation get's rounded).

Another way to see the problem could be (as I think of lossy
compression): given a string consisting of 1's and 0's, in which the
probability of a 1 if 30%, if the entropy of the given string is x,
and you set the wanted size to x/2, what would be the reliability of
the string (i.e. what would be the number of 1's encountered)?

Is there a math process that given those numbers it returns a function
that generates that sequence? Eventually, the sequence can lose some
information in the process (the noise of the data). I wanted to know
if the size to transmit this sequence is lesser than the entropy of
the string itself.

If I wanted manually induce errors, I could, by, let's say, reducing
by half the number of 1's (transmitting only one at each two). But
that doesn't solve the problem that I wanted to solve (finding a
function to maximize the information).

Thanks for the time.

Arsène.


.

Relevant Pages

  • =?iso-8859-1?q?Re:_Kolmorgorov_Complexity_and_Kim_=D8yhus?=
    ... >>of the string. ... These are different forms of potential compression. ... >>functional system may not be able to sustain such sequence compression ... Chaos theory is about DETERMINISTIC systems which amplify small ...
    (talk.origins)
  • Re: =?iso-8859-1?q?Re:_Kolmorgorov_Complexity_and_Kim_=D8yhus?=
    ... There is no compression possible, ... compressed expression is shorter than the expressed string), ... makes no sense to ask whether the DNA-base sequence is shorter than the ... > language system code needed to compress and decompress the sequence. ...
    (talk.origins)
  • =?iso-8859-1?q?Re:_Kolmorgorov_Complexity_and_Kim_=D8yhus?=
    ... >>> erm, if your string is minimum, then it is compressed for the reference ... The sequence of an otherwise functional string may be ... >> compressed without regard to its function, but this compression will ... When Sean says that his definition of functional complexity includes ...
    (talk.origins)
  • Re: Attention Sean - question about CSI
    ... Maybe the degree of compression can serve as a measure ... If I am allowed to choose the compression algorithm *after* ... which compresses that string to a single bit. ... Yet, for a truly random sequence of numbers, you won't find any ...
    (talk.origins)
  • Re: =?iso-8859-1?q?Re:_Kolmorgorov_Complexity_and_Kim_=D8yhus?=
    ... >>>of the string. ... >> universal computing device, say an universal Turing machine, which IS ... K.C. is defined to require an universal computer, ... >These are different forms of potential compression. ...
    (talk.origins)