A Study On Brain-Computer Interface Game Based On Deep Convolutional Neural Network And Its Application

The Brain-Computer Interface(BCI)provides humans with an unconventional communication method and has extremely important application value in many fields.In addition to helping to develop products that help people with disabilities,brain-computer interface(BCI)technology can also provide people with a novel way of entertainment.This kind of entertainment allows people to get rid of the complete dependence on intermediate devices(mouse,keyboard,gamepad and game controller).The previous research mainly focused on applying the existing BCI technology directly to the game,while ignoring the entertainment of the game.Generally speaking,a mature game requires the operation process to be as simple as possible,the user interface is also simple,and the user's commands must be responded quickly.In addition,the game should run normally in an uncontrolled environment,and the user community should be wider.Due to the limitations of poor control performance,fatigue,and poor user experience,most BCI games have not been widely promoted.In order to make BCI games more humane and user-friendly,the strategy of combining control methods and games is still Needs further improvement.Convolutional Neural Network(CNN)is a very popular pattern recognition algorithm.There are also many application prospects in the field of EEG analysis.However,due to the very limited EEG signal data,the problem of overfitting on small training samples is still not well solved.In response to the above problems,this article mainly carried out the following work:(1)We propose a novel simplified Bayesian Convolutional Neural Network(SBCNN)algorithm.The algorithm introduces uncertainty estimation into the model by taking the probability distribution as a parameter,thereby enhancing the robustness of the model.Achieve high accuracy on limited training samples.(2)In this paper,we propose a P300 brain-computer-interface game(MindGomoku)to explore a feasible and natural way to play games by using electroencephalogram(EEG)signals in a practical environment.The novelty of this research is reflected in integrating the characteristics of game rules and the BCI system when designing BCI games and paradigms.(3)To prove the reliability of the proposed algorithm and system control,the algorithm was evaluated on the open data set of the brain-computer interface competition,and 10 healthy subjects were invited to participate in two online control experiments.The experimental results showed that the SBCNN algorithm achieved an average classification accuracy rate of 96%on the public data set.In online experiments,all subjects successfully completed the game control with an average accuracy rate of 90.7%,and the average game duration was more than 11 minutes.These results fully proved the stability and effectiveness of the proposed algorithm and system.This BCI system not only provided users with a way of entertainment but also provided more possibilities for games.