Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Tenchusatsu who wrote (246460)	8/18/2005 2:00:04 PM
From: neolib	Read Replies (1) \| Respond to of 1571911

Couple of problems. One, what sort of "genes" lead to stimuli that reproduce very focused test cases? And how should "genes" be modeled in order to get us to that focused case in the first place? And two, evolution depends on mutation, but the vast majority of mutations are useless, so once again, how long until we produce a mutation that is actually useful and can "survive" for the future If you have a very focused test case, then it would appear that you know what test vector is needed, hence GA would be a step back, not forward IMO. You would simply write the specific test instead. Were GA may help is if you don't know what vector or test is need, but you do have some measure of test result that can be used as a fitness selection. A very simple example is template matching. Say you have a binary string that you cannot observe of N-bits. You can put in an N-bit string and get back a fractional number that reflects a goodness of fit between your input and the unobserved string (with 0 = no match, 1 = complete match). GA algorithms will be much better at finding a solution than simply randomly generating N-bit strings and waiting until you happen to get 1 back. The reason they are better is that they make use of the intermediate information obtained even when the full result is not yet available. A google search for "logic verification" +"genetic algorithm" returned a number of hits from a 1995 DAC conference paper on Power PC & Darwin. You work at IBM?