X-Andrew-Authenticated-as: 7997;andrew.cmu.edu;Ted Anderson
Received: from hogtown.andrew.cmu.edu via trymail for +dist+/afs/andrew.cmu.edu/usr11/tm2b/space/space.dl@andrew.cmu.edu (->+dist+/afs/andrew.cmu.edu/usr11/tm2b/space/space.dl) (->ota+space.digests)
ID </afs/andrew.cmu.edu/usr1/ota/Mailbox/scNl5ES00WBw4AdE4l>;
Tue, 25 Jun 91 05:48:00 -0400 (EDT)
Message-ID: <scNl58O00WBwMAbU5X@andrew.cmu.edu>
Precedence: junk
Reply-To: space+@Andrew.CMU.EDU
From: space-request+@Andrew.CMU.EDU
To: space+@Andrew.CMU.EDU
Date: Tue, 25 Jun 91 05:47:53 -0400 (EDT)
Subject: SPACE Digest V13 #710
SPACE Digest Volume 13 : Issue 710
Today's Topics:
SPACE Digest V13 #614
Re: L-5 Society is now National Space Society (NSS)
Re: Self-sustaining infrastructures
Re: Microgravity?
Re: Beanstalk analysis reprise
Re: INFO: Clandestine Mars Observer Launch??
Astro-Nugget worth $$$ trillions
Re: Why would I stick a face on Mars...
Administrivia:
Submissions to the SPACE Digest/sci.space should be mailed to
space+@andrew.cmu.edu. Other mail, esp. [un]subscription requests,
should be sent to space-request+@andrew.cmu.edu, or, if urgent, to
Just FYI. Certainly seems like a good reason to get out there fast!
From the Washington Post, June 7, 1991 (y114#184), p. A11:
Nearby Asteroid Worth a Trillion
" An astronimical El Dorado containing some 10,000 tons of
gold and 100,000 tons of platinum has been found orbiting
the sun tantalizingly close to Earth, according to a report
in today's issue of Science.
Asteroid 1986 DA, as the solid metal ''near-Earth-object'' is
known is 1.2 miles wide and shaped roughly like a canned ham.
The itinerant astro-nugget also contains approximately 10 billion
tons of iron and a billion tons of nickel.
At today's prices, the commodity value of 1986 DA is approximately
astronomical: The gold alone is worth about $120 billion and the
platinum nearly a trillion dollars.
Immediate prospects for mining, however, are dim: The nearest
the asteroid gets to Tiffany's is about 20 million miles.
Nonetheless, according to astronomer Steven Ostro, who helped
discover the orbiting ingot, ''if in the next century we go into
space robitically or in person, this asteroid -- or many others
yet to be discovered -- could be among the targets of such
missions, and the metals could be used for contruction in space.''
Current theory indicates that such extraordinary metallic
abundances can only form in the cores of planet-like bodies; and
the researchers report in Science that 1986 DA appears to be
''derived from the interior of a much larger object'' that existed
billions of years ago and ''subsequently was disrupted by a
massive collision.''
--
Mike Santangelo (mike@socrates.umd.edu)
UNIX / VMS Systems Manager
------------------------------
Date: 7 Jun 91 21:51:28 GMT
From: cis.ohio-state.edu!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!caen!uwm.edu!csd4.csd.uwm.edu!markh@ucbvax.Berkeley.EDU (Mark William Hopkins)
Subject: Re: Why would I stick a face on Mars...
In article <cHc635w164w@cellar.UUCP> revpk@cellar.UUCP (Brian 'Rev P-K' Siano) writes:
> But what I'd be testing for is this' if something looks like a face,
>does this mean that a face was designed into it?
But you can't even test that with Mount Rushmore this way if you don't know
where it came from?
The point of the test is that the neural net ALREADY is trained to rule out
random images from actual photographs with 99% accuracy. My claim is that
ANY such neural net, no matter how it was designed, trained/programmed when
applied to a photo of the Mars facial image will classify it as a face with
probability exceeding 99%.
You'd be hard pressed to find a random image that would fall between the
cracks then of such a classifier.
You can even use this as the basis for a set of parameters for randomness:
e.g. this image is 99%/99% a face ... that image is 99.9%/99.9% a face, and
so on ... meaning NOT random.
> I think it would amke far more sense to provide subjects with a
>series of patterns, some of which have faces 'designed' into them, and others
>which are truly random... and ask people if they see any faces. I expect that
>one would find a high correlation among the random images of where 'faces'
>could be seen. If people are recognizing faces in verifiably random patterns,
>then our seeing a 'face' on Mars doesn't constitute proof for design.
That's essentially the condition to the test I described: that the neural net
have 99% accuracy. You need a VALID classifier in order to do the test with,