Bat chat: machine learning algorithms provide translations for bat squeaks

They've already done it for dogs:

http://languagelog.ldc.upenn.edu/myl/llog/FarsideDogTranslat...

I heard that an officer by the name of Captain Nathan Bridger, of the The United Earth Oceans Organization, is conducting an animal protocol on board his vessel.

1) You have to train the machine with the known solutions. From the article:

"The team spent 75 days continuously recording both audio and video footage of 22 bats that were split into two groups and housed in separate cages. By studying the video footage, the researchers were able to unpick which bats were arguing each other, the outcome of each row, and sort the squabbles into four different bones of contention: sleep, food, perching position and unwanted mating attempts.

The team then trained the machine learning algorithm with around 15,000 bat calls from seven adult females, each categorised using information gleaned from the video footage, before testing the system’s accuracy."

2) https://en.wikipedia.org/wiki/Decipherment_of_Egyptian_hiero...

"The decipherment of Egyptian hieroglyphs was gradually achieved during the early 19th century."

I'm not sure that the AI significantly under-performed humans. It looks to me like the labels that the accuracy number came from were labeled by watching video of the bats. The 61% understanding number was off of 7 possible topics and averaged over each topic so it's definitely better than guessing. I suspect there's a fair amount of ambiguity and mislabeling in these "topics" from humans trying to interpret bat motivations so 100% accuracy probably isn't really feasible.

I have no idea what the state of the art is in bat understanding but the results seems really impressive to me - maybe I'm easily impressed? :)

I agree the performance isn't too impressive, but it could still be an improvement over trained humans. For example, from the article there's "We have shown that a big bulk of bat vocalisations that previously were thought to all mean the same thing, something like ‘get out of here!’ actually contain a lot of information."

Given that there are four categories of dispute [0] 61% is significantly better than the uniform prior.

[0] "By studying the video footage, the researchers were able to unpick which bats were arguing each other, the outcome of each row, and sort the squabbles into four different bones of contention: sleep, food, perching position and unwanted mating attempts."

It depends. How long did it take the AI to come to the same conclusion we have? Speed of learning is very important.

Ultimate goal: Decipher what women really mean when they speak.