Dolphins communicate through a range of unique clicks and whistles, and now scientists have created an algorithm that can decipher these calls. Pictured is a 3D rendering of the spectrum of clicks

Algorithm can identify distinct click patterns in dolphins

Dolphins communicate through a range of unique clicks and whistles, and now scientists have created an algorithm that can decipher these calls.

The algorithm can identify distinct click patterns among millions of clicks made by wild dolphins.

Researchers hope the technology could be used to help distinguish between dolphin species in the wild.

Scroll down for video

Dolphins communicate through a range of unique clicks and whistles, and now scientists have created an algorithm that can decipher these calls. Pictured is a 3D rendering of the spectrum of clicks

Dolphins communicate through a range of unique clicks and whistles, and now scientists have created an algorithm that can decipher these calls. Pictured is a 3D rendering of the spectrum of clicks

HOW DOES IT WORK?

The researchers developed an algorithm that can pick out consistent click patterns in very large datasets.

Rather than being taught to recognise patterns that are already known, the algorithm is ‘unsupervised’, and seeks patterns on its own.

During testing, the algorithm was able to identify consistent patterns in a dataset of over 50 million clicks recorded in the Gulf of Mexico over a two-year period.

These click types were consistent across monitoring sites in different regions of the Gulf, and one of the click types that emerged is known to be associated with a dolphin species.

Researchers from the Scripps Institution of Oceanography in San Diego build autonomous underwater acoustic sensors that can record dolphins’ clicks in the wild for over a year at a time.

These instruments are used to study a range of aspects of dolphin populations, including how they are affected by the Deepwater Horizon oil spill, natural resource development, and climate change.

But because the sensors record millions of clicks, it’s difficult for a human to recognise any species-specific patterns in the recordings.

To overcome this issue, the researchers developed an algorithm that can pick out consistent click patterns in very large datasets.

Rather than being taught to recognise patterns that are already known, the algorithm is ‘unsupervised’, and seeks patterns on its own.

During testing, the algorithm was able to identify consistent patterns in a dataset of over 50 million clicks recorded in the Gulf of Mexico over a two-year period.

These click types were consistent across monitoring sites in different regions of the Gulf, and one of the click types that emerged is known to be associated with a dolphin species.

The team suggests that the algorithm could help to match other dolphin species, and monitor wild dolphins. This would be a vast improvement on current systems, which rely on people observing the dolphins from ships or aircraft (stock image)

The team suggests that the algorithm could help to match other dolphin species, and monitor wild dolphins. This would be a vast improvement on current systems, which rely on people observing the dolphins from ships or aircraft (stock image)

WHAT IS ECHOLOCATION?

Dolphins use a process called ‘echolocation’ to help them identify where objects are located.

The animals send out clicking noises in the form of sound waves, which hit an object and bounce back vibrations.

The amount of time it takes for the sound waves to come back helps the dolphins to identify their distance from the object – the longer the sound waves, the further away the object.

It also gives them information about the location of the object and some indication of the shape and size of it.

The team suggests that the algorithm could help to match other dolphin species, and monitor wild dolphins.

This would be a vast improvement on current systems, which rely on people observing the dolphins from ships or aircraft.

The researchers now plan to integrate the algorithm with deep learning methods to improve its ability to identify click types.

Dr Kaitlin Frasier, lead-author of the study, said: ‘It’s fun to think about how the machine learning algorithms used to suggest music or social media friends to people could be re-interpreted to help with ecological research challenges.

‘Innovations in sensor technologies have opened the floodgates in terms of data about the natural world, and there is a lot of room for creativity right now in ecological data analysis.’

Posted on; DailyMail>>

Check Also

Earth

Revealed: Scientists uncover MYSTERY deep underwater sound that might be the hum of Earth



SCIENTISTS have recorded a mystery sound from the depth of the ocean, which could hold the answers to the Earth’s makeup.

Leave a Reply

Your email address will not be published. Required fields are marked *