What is Brainsourcing and Its Applications
What is Brainsourcing?
Brainsourcing is a technique to train a machine learning model by using a group of human contributors to perform a recognition task. The trained model or the classifier can be used to classify images from predicting the input stimulus. In this way, the brainsourcing model performs better than an individual model with higher accuracy.
Where does the idea of brainsourcing come from?
Crowdsourcing is a paradigm in which a big task is distributed among a group of people to achieve a collective result. It could further be classified into macrowork and microwork. Macrowork refers to the work that requires people with specialized skills, and it takes a longer time to complete. Microwork, on the other hand, is the work that does not need specialized skills to complete within a short amount of time. When unifying the fields of crowdsourcing and brain-computer interfacing, a novel technique, Brainsourcing, is created to speed up and increase the accuracy of machine learning processing of image recognition.
How does Brainsourcing work in the brain-computer interfacing?
The electroencephalography, also known as EEG, is a popular means of recording the brain's electrical activity. Any muscle movements or thinking activities in the brain, measurable changes in voltage at the scalp's surface. However, EEG gives better time-related information or the temporal resolution but is unable to provide clear data about spatial information or the location of the stimuli. A derivative of EEG, Event-related potentials (ERPs), is used instead since it gives a better spatial resolution. For example, moving a muscle or viewing an image, ERPs are produced by the brain in response to the event. ERPs can further characterize by its positive and negative components from its wave presentation. These positive and negative differences give crucial data for machine learning to distinguish the response of an event. Specifically, the component is called the P300 (p3) wave in neuroscience. In the 1980s, the use of P300 ERP was widely used for a lie detection application.
Brainsourcing achieves results by combining derived from many individuals' brains, in contrast to a single individual. One must understand that everyone's stimuli wave is unique, just like a fingerprint. So although every participant saw the exact image, everyone's response is different with some degrees of difference. As a result, when combining more sample stimuli, a better training model can be achieved.
Practical Applications to Brainsourcing Machine Learning Method and Its Issues
One application of Brainsourcing is to identify human preferences by concluding using groups of people's brain activity. Other potential applications of this new technique would be many. For the film or marketing industry, they may use this technology to find the optimum setting of a scene or an image of representing the emotion that they want to stimulate and deliver the intended results. For better TV rating reporting companies like Nielsen, they may want to find out exactly how people like the show from paid viewers tapping their brain activity.
Due to the brain activity being unique like a fingerprint, there could be ethical concerns if there are ill intentions. The brainsourcing workers who participate in the brainsourcing task could give away their unique signature, and their rights to privacy or autonomy could be violated. Governments, corporations, and criminal organizations could use this technology to gain financial gain or limit mental and physical activities. Legal ethical standards need to establish, and moral values must be maintained to govern technology advancement.
Reference: https://dl.acm.org/doi/abs/10.1145/3313831.3376288