Re: Ping: dawn, some mvl questions

David Cressey wrote:

> "Keith H Duggar" <duggar_at_alum.mit.edu> wrote in message
> news:1148249698.686154.221010_at_i40g2000cwc.googlegroups.com...
>

>>dawn wrote:

>>>information never verbalized
>>
>>(though "verbalized" seems a poor choice of words perhaps
>>for "modeled" or "represented"?)

Noise is information. I am reminded of an article I read long ago. I think it was perhaps one of the first Byte magazines I read--back when the magazine was good and stuff.

The article explained some aspects of image processing. How NASA found it necessary to invent image processing to 'sell' the photographic and video results they got back from various space missions etc.

The article showed the effects of a median filter to remove thermal noise. The 'before' image showed a very snowy picture of some white flowers on a dark background. The 'after' image clearly showed highly detailed pictures of some trilliums growing in a forest near a log. One could see the texture of the log, the leaves of the surrounding trees and details in the petals.

The article made the point that the 'after' image had less information than the original. If you remove more noise than signal, the signal to noise ratio improves. But the result has less information.

Had the picture shown mayflies and blackflies instead of thermal noise, the clear image would remove all information about the insects.

Since data is information encoded for automated processing, a loss of data is a loss of information. However, not all information is equally valuable.

   I would contend that there are sitations where data
> can be discarded without losibg information, provided its information about
> some specific context.

Your contention would be false. Had you contended, instead, that there are situations where one can lose information without losing anything of value, your contention would have been true. Received on Mon May 22 2006 - 22:10:28 CEST