Re: Effective Lossy Compression of Bitstream


Hi, Thomas.

Simplifying, I have a bitstring of ones and zeros, which the
probability of a one is about 30%.
I want to compress lossy this bitstring, and the process can erase
some 1's, but it cannot add any ones.

IOW, you want to generate a sequence such that the conditional entropy of
the generated sequence given the original is maximal, and the entropy of
the generated sequence is bounded by a threshold?

IOW, yes. I wanted a function to "transform" the original sequence, in
a way that the new sequence could be better compressed, while
retaining as much reliability (as I measure by the relation of the
number of ones in the original bitstring and the compressed one) as
possible.

As an example, if the most ones follow a exponential function (ex:
x^2), then I could transmit the function to recreate the sequence. Of
course maybe not all '1's of the original could be recovered, but if
the size to transmit the function is small, then the gain would
justify it's use.

Really sorry for not making myself much clear, to me it seemed an easy
problem.

Arsène.
.

Relevant Pages

  • Re: Best existing binary compressor method?
    ... The best way to encode this source is in zero bits. ... an encoder design *never* compresses a specific sequence. ... compress below this entropy. ... Don't transmit anything, we know the sequence, problem solved. ...
    (comp.compression)
  • Re: Probability, compressible sequences: Backgrounds
    ... noise is caused simply because people do not use adequate definitions ... A "specific sequence" cannot be random. ... How can you compress a specific sequence wthin the decompressor to ... external trigger (me, dropping the coin). ...
    (comp.compression)
  • Re: Compression leads to encryption NEW COMPRESSION METHOD!
    ... Can you please present a simple pseudo code example of your alorithm? ... Looking at your PPT, it seems you have a simple substitution system inplace. ... in a 16 bit sequence, and replace it with the most compressible (via my ... still get 5 to 8 cycles before we can no longer compress, ...
    (sci.crypt)
  • Re: Probability, compressible sequences: Backgrounds
    ... In the compressed sequence you need to insert also the decompressor ... How can you compress a specific sequence wthin the decompressor to ... I'm not using the coin experiment to define what a random process is. ... the only reason why we use compressors is that we model data ...
    (comp.compression)
  • Re: JSH: The will to lie
    ... is more likely to be followed by a residue of 2> ... The obvious thing is to code in binary, and then compress. ... Isn't the obvious compression algorithm for this sequence the one you ... The sequence is given by a simple algorithm. ...
    (sci.math)
Loading