Yannis Assael at DeepMind and colleagues trained an artificial intelligence (AI) algorithm to guess missing words or characters from Greek inscriptions up to 2,600 years old.
The Pythia algorithm learned to recognize patterns on 35,000 relics containing more than 3 million words.
The patterns it identifies include the context in which different words appear, the grammar, and the shape and layout of the inscriptions.
In a head-to-head test on the missing parts of 2,949 damaged inscriptions, human experts took two hours to get through 50 inscriptions, while Pythia gave its answers in seconds, and was 30% more accurate.
Given an inscription with missing information, Pythia provides 20 different suggestions so experts can then select the best one.
From New Scientist
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA