Recommendation algorithms used by Amazon and Netflix improve satellite imagery

The authors of the new paper found that recommendation algorithms that help consumers decide what

watch or buy, can help satellites better study the Earth.

Optical satellites stop fixing the Earth,when it's covered in clouds. Therefore, researchers have long relied on inaccurate instruments to study blind spots, especially along coastlines. The authors of the new paper adapted the recommendation algorithm that Netflix uses to create a system that predicts coastal cloudscapes more accurately and quickly than conventional data-filling tools.

The authors note that the way that recommender systems process data is not much different from the process of predicting coastal landscapes hidden by clouds.

To test their hypothesis, the authors builtcloud filling model. An algorithm called Funk-SVD - originally it displayed consumer reviews on a matrix. This data is then used to predict content that may be of interest to the user. 

A similar process was used to fillclouds: each coordinate on the map is represented by a pixel in the photo, and that pixel can be either water or land, and clouds represent unregistered data. Wang's adaptation of Funk-SVD allows one to make guesses about what is under the clouds based on other data.

Read more:

American satellite "saw" an unusual message from Earth

A video has been published from a rocket launched from an experimental acceleratorGiant funnel found in China. Species unknown to science may be hiding there.