Two thousand year old scroll decoded using machine learning

[colonel_mortis]

Summary

 

Several paragraphs of text have been decoded from the inside of a scroll that was buried during the eruption of Mount Vesuvius in AD 79. These scrolls were too fragile to physically unroll, so researchers took high resolution CT scans of the scrolls, and released the data to the public, promising $700,000 to the first team to decode 4 passages from the inside of the scroll, based on the scan (along with a number of other prizes along the way), before the end of 2023.

 

Quotes

Quote

There was one submission that stood out clearly from the rest. Working independently, each member of our team of papyrologists recovered more text from this submission than any other. Remarkably, the entry achieved the criteria we set when announcing the Vesuvius Challenge in March: 4 passages of 140 characters each, with at least 85% of characters recoverable. This was not a given: most of us on the organizing team assigned a less than 30% probability of success when we announced these criteria! And in addition, the submission includes another 11 (!) columns of text — more than 2000 characters total.

 

The results of this review were clear and unanimous: the Vesuvius Challenge Grand Prize of $700,000 is awarded to a team of three for their excellent submission. Congratulations to Youssef Nader, Luke Farritor, and Julian Schilliger!

 

 

The submission contains results from three different model architectures, each supporting the findings of the others, with the strongest images often coming from a TimeSformer-based model. Multiple measures prevent overfitting and hallucination, including results from multiple architectures, a study across input/output window sizes, label smoothing, and varying validation folds. Like with all our prizes, this ink detection code has been made public as open source (on GitHub), leveling up everyone in the community.

Quote

What does the scroll say?

 

To date, our efforts have managed to unroll and read about 5% of the first scroll. Our eminent team of papyrologists has been hard at work and has achieved a preliminary transcription of all the revealed columns. We now know that this scroll is not a duplicate of an existing work; it contains never-before-seen text from antiquity. The papyrology team are preparing to deliver a comprehensive study as soon as they can. You all gave them a lot of work to do! Initial readings already provide glimpses into this philosophical text. From our scholars:

The general subject of the text is pleasure, which, properly understood, is the highest good in Epicurean philosophy. In these two snippets from two consecutive columns of the scroll, the author is concerned with whether and how the availability of goods, such as food, can affect the pleasure which they provide.

Do things that are available in lesser quantities afford more pleasure than those available in abundance? Our author thinks not: “as too in the case of food, we do not right away believe things that are scarce to be absolutely more pleasant than those which are abundant.” However, is it easier for us naturally to do without things that are plentiful? “Such questions will be considered frequently.”

Since this is the end of a scroll, this phrasing may suggest that more is coming in subsequent books of the same work. At the beginning of the first text, a certain Xenophantos is mentioned, perhaps the same man — presumably a musician — also mentioned by Philodemus in his work On Music.

Philodemus, of the Epicurean school, is thought to have been the philosopher-in-residence of the villa, working in the small library in which the scrolls were found.

Quote

Vesuvius Challenge Stage 2

In 2023 we got from 0% to 5% of a scroll. In 2024 our goal is to go from 5% of one scroll, to 90% of all four scrolls we have scanned, and to lay the foundation to read all 800 scrolls.

The primary goal for 2024 is to read 90% of the scrolls, and we will issue the 2024 Grand Prize to the first team that is able to do this. More details on the exact grand prize judging criteria will be available in March.

 

My thoughts

This is an achievement that could only have been done with machine learning, and the technical feat here can't be understated. It has taken a lot of work to get this far, it's incredible to see what the community can achieve when it's given a goal like this. The $1M+ prize pool (donated by various mostly rich people) certainly helped to incentivise people to participate, and it will be interesting to see if this model gets adopted for any other projects in the future. If they are able to achieve their goal of extending this technology to read all 800 scrolls, this will be a big breakthrough in our understanding of Ancient Rome, let alone the other potential places where this technology could be used.

 

Sources

Official announcement: https://scrollprize.org/grandprize

Edited February 6 by colonel_mortis
Add details about what the scroll says

[filpo]

is it just me or does the scroll look a bit like shit? (literally)

[OhYou_]

well, what does the scroll say?

[TheLANguy]

Really impressive! The pictures of the swirly mess that the researchers started with really shows how impressive this is!

14 minutes ago, OhYou_ said:

well, what does the scroll say?

It says on the scroll website. Just scroll down a little.

https://scrollprize.org/grandprize

[Zomeguy]

20 minutes ago, filpo said:

is it just me or does the scroll look a bit like shit? (literally)

I was thinking mummified finger.

[manikyath]

Quote

Multiple measures prevent overfitting and hallucination, including results from multiple architectures, a study across input/output window sizes, label smoothing, and varying validation folds.

if only all AI projects had this much courtesy to double-check results.

[da na]

Amazing, it's philosopher beef.

 

I wonder if people 2000 years from now will be this excited about digging up harddrives with twitter arguments

[Godlygamer23]

1 hour ago, da na said:

Amazing, it's philosopher beef.

 

I wonder if people 2000 years from now will be this excited about digging up harddrives with twitter arguments

I would still say that it's a fascinating and humorous look into our past lol. I myself find it fascinating looking back even within my current lifetime to see how things have come, especially in light of how my understanding has changed over the years with various topics. 

 

Sometimes older people find my fascinating offensive, almost like I'm comparing them to cavemen, but in reality, it's a genuine fascination with an era that I wasn't part of. 

[filpo]

10 hours ago, TheLANguy said:

scroll website. Just scroll down a little.

[FlyingPotato_is_taken]

Can we appreciate how amazing CT have gotten? Being able to resolve 5µm thick layer of ink rolled up into a scroll.

[Mark Kaine]

8 hours ago, FlyingPotato_is_taken said:

Can we appreciate how amazing CT have gotten? Being able to resolve 5µm thick layer of ink rolled up into a scroll.

i agree, that's the impressive feat here, how the computer algo is called that they used to make sense of it seems less relevant. 

[Kisai]

On 2/6/2024 at 1:19 PM, colonel_mortis said:

 

My thoughts

This is an achievement that could only have been done with machine learning, and the technical feat here can't be understated. It has taken a lot of work to get this far, it's incredible to see what the community can achieve when it's given a goal like this. The $1M+ prize pool (donated by various mostly rich people) certainly helped to incentivise people to participate, and it will be interesting to see if this model gets adopted for any other projects in the future. If they are able to achieve their goal of extending this technology to read all 800 scrolls, this will be a big breakthrough in our understanding of Ancient Rome, let alone the other potential places where this technology could be used.

 

Sources

Official announcement: https://scrollprize.org/grandprize

This is what AI should be focusing on, translation/transcription, because that is most helpful for us. Any form of accessibility enabled by AI is plus. There is not enough people in the world that would be willing to learn some obscure language just for the sake of translating and transcribing some long dead documents.

[hu4d]

On 2/6/2024 at 5:22 PM, manikyath said:

if only all AI projects had this much courtesy to double-check results.

if only i had the mental capability to understand what they even said lol

[StDragon]

1 hour ago, hu4d said:

if only i had the mental capability to understand what they even said lol

It's a message from Doc Brown. He needs yet another time travel lift from Marty.