US military uses soldiers' brains as a neural network for image analysis
At the MIND military research laboratory (Mission Impact through Neurotechnology Design, or completing tasks through neurotechnology), specialists work on technology for accelerated image analysis, processing signals from the human brain. The military believes that in this way it will be possible to efficiently process large amounts of information.
The task of computer vision in general and analysis of photographs in particular, information technology specialists are using artificial neural networks, approximately simulating the work of the human brain. However, the army decided to do differently - instead of building computer neural networks and the complex process of training them, use what the army has in abundance. Namely, a soldier.
The human brain is better than any artificial neural network to cope with image recognition. Moreover, computer networks need to be “trained” for a long time on specially selected sets of images. For this, a person trains himself and his whole life. Why not use his "computing power" for his usual tasks?
To test the concept, the experimental military was asked to choose one of several items (boat, panda, strawberry, butterfly, chandelier), and not to inform about their choice. Then, with a frequency of approximately once per second, the researchers showed them photographs of various objects, taking readings of an electroencephalogram.
The employee did not need to react in any way to the appearance of the image - it was enough to count their number in the mind. When a photograph of an object conceived by the subject appeared on the screen, the device read out an obvious signal that could be unambiguously interpreted. After passing the two-minute test, it was possible to say with 100% accuracy which of the five subjects was conceived by the experimental.
In the second test, to test the possibilities of accelerating the process, the subjects were shown only part of the photograph, but about five of these parts were shown per second. And again, according to the EEG readings, it was possible to clearly see at what point the brain recognizes the conceived object by a fragment of the image.
Dr. Anthony Ries, a specialist in cognitive neuroscience, explains the idea. The fact is that the ability to collect information, ranging from satellites to drones, has already exceeded the ability to process received data.
When an analyst studies a large aerial photograph in search of the desired object, it takes a lot of time. If the military manages to apply the developed technology, then it will be possible to break a large image into small fragments, quickly “scroll” them before the eyes of analysts, and use the data from the EEG to immediately process the results.
Then, instead of gradually studying the map, making notes or making any marks, wasting valuable time, the analyst will only have to look at the screen, focusing on the object of interest (convoy, caravan, warship, etc.).
Now that the concept has gained the right to exist, the next step in research is the elimination of spurious EEG signals, which can occur due to various psychological and physiological factors inherent in a living organism.