Neural network Ithaca will help archaeologists read ancient texts

(ORDO NEWS) — The neural network and machine learning find application in everything. It became known that a subsidiary of Alphabet DeepMind has developed an artificial intelligence capable of understanding and restoring text on ancient artifacts.

It is also reported that the neural network, called Ithaca, is even able to determine the period when this text was written.

As the developers of the neural network explained, the intelligence is trained on almost 80 thousand examples of ancient Greek artifacts.

In addition, each inscription is endowed with metadata when a particular entry was made. On the basis of the introductions, Ithaca is already forming its own idea of ​​the missing signs and date.

However, as the authors of the development explained, the neural network does not guarantee 100% accuracy of reproduction of the original text. For this reason, Ithaca is not a stand-alone software for determining the history of a find.

The neural network, roughly speaking, is advisory in nature: Ithaca will tell you what could be in place of the missing signs, but the correctness and consistency of the result will still be checked by a person.

In the future, the developers plan to increase the “localization” of the neural network, as well as improve recognition accuracy by introducing more text samples.


Contact us: [email protected]

Our Standards, Terms of Use: Standard Terms And Conditions.