[seqfan] Re: Graphical recognition software, anyone?

[Simon Plouffe]

hello Neil,

  facial recognition uses eigenvalues
and some matrices calculation. Perhaps
an algorithm can be used to classify those
300000 graphics.

  See here : https://en.wikipedia.org/wiki/Eigenface

Well, I will think of something to do the job,

  ... my first impression is : we do not need something
so sophisticated,

  Best regards,
  Simon


Le 2022-02-12 à 18:46, Neil Sloane a écrit :
> Dear Sequence Fans:
> If you go to a sequence in the OEIS and click "graph" you see two plots.
> The lower one is a scatter plot that typically shows you 10K terms.
> Could we use facial recognition software to classify these plots, and in
> particular to look for the best match to one we are trying to analyse?
>
> The plots of existing sequences would give us a database of say 100K plots.
>
> (Our plotting software - which uses R - automatically produces a log-plot
> if that seems appropriate. But that's fine, we just look for a match to the
> log-plot.  If the sequence has a b-file, typically 10K or 20K terms, that
> is used for the plot.)
>
> For example, I was just looking at these 4 related sequences,
> A267099, A351168, A351430, A351436.  It would be nice to be able to call up
> a program that would simply look at all the plots of the OEIS entries, and
> find the closest matches to them.
>
> Can we turn spyware into sequence-ware?
>
> Best regards
> Neil
>
> Neil J. A. Sloane, Chairman, OEIS Foundation.
> Also Visiting Scientist, Math. Dept., Rutgers University,
> Email:njasloane at gmail.com
>
> --
> Seqfan Mailing list -http://list.seqfan.eu/