How is Google's open source machine learning system "TensorFlow" used?

Machine learning system that Google uses for product development, commercial use OK and open source is also being done "TensorFlow"Is used not only by specialized researchers but also by general high school students as a machine learning system that anyone in the world can use easily. The case is introduced by Google on blog.

How TensorFlow is powering technology around the world
https://www.blog.google/topics/machine-learning/how-tensorflow-powering-technology-around-world/

At the University of PennsylvaniaAri SilburtHe is trying to solve the origin of the solar system by machine learning using TensorFlow. Mr. Silburt exists in the stars in the solar systemcraterIt is said that they are investigating where the substance exists in the solar system.

Traditionally, we mapped craters manually from photos of stars, but this method requires a huge amount of time. So, Mr. Silburt used TensorFlow to train the machine learning model to learn the data of the existing moon and succeeded in making the crater automatically discriminate from the photograph. In the following image, the left side shows the picture showing the craters and the right side shows the craters judged from the picture.


Perform activities to conserve Brazilian rainforestRainforest ConnectionFounder ofTopher WhiteHe said that he developed an alarm device that prevents illegal tropical rainforest logging in the Amazon using TensorFlow. This warning device reusing mobile phones that are no longer used is a system installed in trees of tropical rainforests to recognize the sound of chainsaws and logger's trucks and to contact the rangers who are alarmed in the area I will. Until now, people seemed to have patrolled warning areas and were looking for loggers, but by introducing this system, we were able to watch the tropical rainforests even without humans.


It is one of the complications of diabetesDiabetic retinopathyIs a terrible illness that progresses at a very fast speed and leads to blindness and the risk of developing more than 400 million people worldwide. In order to discover this disease that can be treated if it can be found at an early stage, Google announced that diagnostic methods using machine learning are being utilized in 2016.

patient'sFundusIn addition to being able to detect diabetic retinopathy at an early stage by accurately analyzing photographs,OptometristIsJorge CuadrosDr. says, "By evolving the machine learning model more, it is also possible to judge the risk of cardiovascular disease in patients".


Examples of high school students using TensorFlow are also introduced.Shaza MehdiWith Mr.Nile RavenellMrPlantMDWe developed an application that diagnoses whether plants are suffering from disease. This application using TensorFlow for machine learning system,PlantVillageIt is said that AI has learned data existing in farmer's site for agricultural workers and database of university and developed to discriminate diseased plants. Mr. Mehdi has developed another application to diagnose skin diseases by applying this technology.


TensorFlow which was made open source seems to be utilized in various scenes from the purpose of unraveling the mystery of the universe to the diagnosis of diseases of plants.