Academic Journal of Computing & Information Science, 2024, 7(8); doi: 10.25236/AJCIS.2024.070807.
Yifan Fu1, Zhenni Chen2, Guoqing Li3
1Department of Economics and Finance, University of Sanya, Sanya, 572022, China
2College of Finance and Economics, University of Sanya, Sanya, 572022, China
3Saxo FinTech Business School, University of Sanya, Sanya, 572022, China
This paper investigates several machine learning methods employed in brain-computer interface applications and focuses on their use in addressing overfitting and improving prediction accuracy. Firstly, the concept of L2 regularization and its role in model training are introduced, which effectively reduces the problem of overfitting by controlling the model complexity. Then, the application of convolutional neural network (CNN) in visual image analysis is introduced in detail, emphasizing that it effectively reduces the number of parameters through local receptive field, weight sharing and pooling layer, and alleviates the overfitting of the model. Then the ensemble idea of random forest algorithm and its application in classification and regression are discussed, especially how to use randomness to improve the prediction accuracy in the process of processing multiple decision trees. Finally, the performance of different algorithms on actual data sets is analyzed, and it is pointed out that convolutional neural network has more outstanding performance in recognition accuracy than support vector machine and random forest, especially in complex signal discrimination and individual difference processing.
Convolutional neural network model, recognition accuracy of brain-computer interface, overfitting
Yifan Fu, Zhenni Chen, Guoqing Li. Research on recognition accuracy and overfitting of brain-computer interface based on convolutional neural network model. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 8: 42-48. https://doi.org/10.25236/AJCIS.2024.070807.
[1] Chen Hong, Yan Jianguo, Yang Hua, et al. Based on the parametric structure depth of separable convolutional neural networks [J/OL]. Journal of Beijing university of aeronautics and astronautics, 1-15 [2024-07-05].
[2] Han Shuo, Yuan Weicheng, Du Zeyu. Based on the feedback threshold learning a semi-supervised skin cancer diagnosis model [J/OL]. Journal of hebei university (natural science edition), 1-8 [2024-07-05].
[3] Huang, Y., Lu, L., Shen, H. et al. Invasion of insect identification based on convolution neural network study [J]. Journal of Chinese agricultural mechanization, 2024, (7): 222-227 + 260.
[4] Yu Q W, Zhou W C, Dai Y K, et al. Embedded systems movement imagine brain-computer interface decoding algorithm in [J/OL]. Computer engineering and application, 1-21 [2024-07-05].
[5] Jiang, Y. & Liu, X. Research status and development trend of brain-computer interface technology [J]. Science and Technology Horizon, 2024, 14(04):56-58.
[6] Wang, X., Julie, W. & Kong, W. Z. New brain more exercise to imagine brain-computer interface system [J]. Journal of hangzhou university of electronic science and technology (natural science edition), 2022, and (6): 33-39.
[7] Xu M P, Luo R X, Xiao X L, et al. Extremely weak brain electrical decoding technology based brain-computer interface front looking [J]. Journal of artificial intelligence, 2021, (6): 101-105.
[8] Chen, Y., Sun, F. & Yao, Y. et al. Research on EEG signal analysis and discrimination method for rehabilitation engineering [J]. Mathematics in Practice and Cognition, 2021, 51(23):197-204.
[9] Guo Dong. brain-machine interface: a lot of imagination to imagine the future [J]. China's Ministry of Industry and Information Technology, 2020, (12): 18 to 23.
[10] Li Z, Wang Yan, Liu F J, et al. De-noising of sleep EEG signal [J]. Chinese Journal of Tissue Engineering Research and Clinical Rehabilitation, 2007, (13):2531-2533.