Message:

<p>[QUOTE=&quot;geekpryde, post: 2013982, member: 36248&quot;]Since Machine Learning is mostly based on inputs (one branch is), coin grading seems like a perfect test case for this technology to me.&nbsp; &nbsp; Basically, you show the machine up 100-1000 examples of each grade for each series&nbsp; (according to [USER=112]@GDJMSP[/USER] of course <img src="styles/default/xenforo/clear.png" class="mceSmilieSprite mceSmilie7" alt=":p" unselectable="on" unselectable="on" />).&nbsp;&nbsp; In some series, this wont be possible, but lets start with the easy ones.&nbsp;&nbsp; Some examples would be toned and not, attractive and not, high-end or low-end for the grade, etc.&nbsp; Basically a great variety for as many series in as many grades as possible, including problem coins of all kinds.&nbsp; &nbsp;&nbsp; You then let the machine use the inputs to create the necessary &quot;mapping&quot; of what each grade looks like, a fancy algorithm that human mind probably cant ever imagine or create, with all the variance that entails, and then you give it another 10,000 coins and see how a Machine grades vs a human.&nbsp; &nbsp; &nbsp;&nbsp; Think computer chess competitions in the 90&#039;s.&nbsp;&nbsp; </p><p><br /></p><p>This isn&#039;t really theoretical, this could be done now to see how consistent machines could grade.&nbsp; &nbsp;&nbsp; I bet, they would at the very least be more consistent than TPG.&nbsp;&nbsp; Of course as other have pointed out, watch out what you wish for.&nbsp; &nbsp;&nbsp; We very well might not like the results of machine grading at all!!![/QUOTE]</p><p><br /></p>

Your name or email address:

Do you already have an account?

• No, create an account now.
• Yes, my password is:
• Forgot your password?

Stay logged in