|
|
On 17/02/2015 11:08 PM, Lars R. wrote:
> Nice long posting. :-D
TY.
> The initiators of that prize also claim that good (lossless) compression
> is strongly related to AI and learning neuronal networks. :-)
Like I said, PAQ uses several statistical models together with an
*artificial neural network* to combine the estimates into a final total.
In other words, the ANN is there to "learn" which model(s) work best for
this specific file.
*sigh* Oh how I haven't missed our GAANN lectures... Mostly because of
the lecturer involved, not the content.
Post a reply to this message
|
|