Violin Forum/Message Board

kubasa · Member Joined: 13 Sep 2007 Posts: 212

Did anyone come across or read this article yet?

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0109229#abstract0

I'll be honest that I really haven't read it. The topic seems interesting, though. I hope to read it this weekend. It was published Oct. 8 so it's fairly new.

James

DonLeister · Posted: Wed Oct 15, 2014 7:03 pm Post subject:

Yes, a friend sent it to me and I skimmed it. I just looked at it again and found the 'conclusion' part luckily. It's a dense piece of number crunching it looks like.

ctviolin

Mat Roop · Posted: Thu Oct 16, 2014 10:41 am Post subject:

Looks like a PHD thesis? I read the conclusion, but for me, there was nothing that struck me as practical or useful.
Cheers, Mat

DonLeister · Posted: Thu Oct 16, 2014 12:14 pm Post subject:

I agree with you CT.
I also wondered about how the writer measured the instruments, photos I presume.
Yah, I was hoping for something practical or useful too Mat!

Janito · Member Joined: 08 Oct 2007 Posts: 114 Location: USA

I think you may be missing an important point.

The author is showing how an algorithm can group violins in clusters based on certain characteristics.

Now assume that you have data on many different violins from many diverse sources, photos, CT scans, measures etc..

Might you be able to generate a "Charles Beare" in silico?

The analysis of Big Data is here, but not yet for violins.

Michael Darnton · Posted: Fri Oct 17, 2014 8:42 am Post subject:

There was someone on the web a few years ago who was "certifying" violins based on photos, matching them via software to a database he'd assembled of the locations of principal points on outlines and f-holes. I seem to remember it was costing a couple of hundred bucks, and he was bragging to have "discovered" a Landolfi violin that didn't look like a Landolfi to me at the time. In general, the idea is a spectacularly bad one to anyone who knows violin making history.
_________________
new blog at my site! http://darntonviolins.com/blog
my work sites: http://darntonviolins.com and http://darntonhersh.com

ctviolin · Posted: Fri Oct 17, 2014 10:55 am Post subject:

Janito · Member Joined: 08 Oct 2007 Posts: 114 Location: USA

Spectacularly bad idea - perhaps.

Missing the point - perhaps.

Now consider this: apparently unconnected data can point to the identity of a unique individual - the sum being much greater than the component parts. This has led to a huge emphasis on the legal handling of "personally identifiable information" collated by trawling systems.

That would happen with violin-related information if there is ever anyone interested enough to fund the exercise.

We should have this discussion again in 10 years time.

Michael Darnton · Posted: Fri Oct 17, 2014 5:10 pm Post subject:

The problem with violins is that the transmission of this data is not organized, or linear, or sensible. Someone in Poland in 1840 could get a Strad in his hands and if he was smart, copy enough details to convince someone who didn't know enough that it was a Strad.

There was a committee of Swiss police "experts" in the 60s or 70s who put together a series of books based on visible and measurable data points, along with UV and infrared. Their goal was to assemble a set of verifiable examples of the great makers to work from--to show the dumb violin makers how forensic experts could sort it with science. Unfortunately, their books were mostly filled with crap, because violins resist brute force attempts to organize them. Some of my violins look more like Strads in many more features than some Strads I have seen, for instance.

The art world has the same problem. There are personal issues much more subtle than machines can pick up. Janito, did you see the Rembrandt and fakes show at the NC Museum of Art? I went and it was wonderful. A mix of real and fake. About 1/3 of the way through, I started to be able to sort them, but I would challenge a machine to do it on the basis that I used.
_________________
new blog at my site! http://darntonviolins.com/blog
my work sites: http://darntonviolins.com and http://darntonhersh.com

Janito · Member Joined: 08 Oct 2007 Posts: 114 Location: USA

ctviolin · Posted: Sun Oct 19, 2014 4:45 pm Post subject:

kubasa · Member Joined: 13 Sep 2007 Posts: 212

Well, I FINALLY sat down and read the article. I guess I didn't walk away with anything like others mentioned. I picked up a lot more from the subsequent postings after mine in this thread than the article!

James