Fully connected network samples transfer and multi-classifier fusion for motor imagery recognition.
Saved in:
| Title: | Fully connected network samples transfer and multi-classifier fusion for motor imagery recognition. |
|---|---|
| Authors: | Cheng, Sihui1 (AUTHOR) chengsihui@stumail.ysu.edu.cn, Gao, Chang1,2 (AUTHOR) gc_ysu@ysu.edu.cn |
| Source: | Neural Computing & Applications. May2025, Vol. 37 Issue 13, p8153-8164. 12p. |
| Subjects: | Motor imagery (Cognition), Marginal distributions, Network performance, Generalization, Sampling methods |
| Abstract: | In the field of motor imagery (MI) recognition, there are two problems, which are poor generalization and low recognition performance. A method of MI recognition method based on fully connected network (FCN) samples transfer and multi-classifier fusion is proposed in this study. The distribution similarity-based source domain selection method seeks source domains with marginal distributions that are similar to the target domain. Then, an FCN-based samples transfer method is proposed for transferring suitable samples from similar source domains to the target domain with similar conditional distributions. Additionally, a method for MI recognition is proposed that is based on the fusion of multiple classifiers. The classifiers are trained on labeled samples from the target domain as well as samples transferred from similar source domains. The new sample in the target domain can be identified using the weight fusion of the results of these classifiers. To evaluate the proposed method's effectiveness, four types of MI from the brain–computer interface competition IV dataset 2A were used to evaluate the recognition ability, and the results confirmed excellent recognition and generalization performance compared to commonly used methods nowadays. [ABSTRACT FROM AUTHOR] |
| Copyright of Neural Computing & Applications is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | In the field of motor imagery (MI) recognition, there are two problems, which are poor generalization and low recognition performance. A method of MI recognition method based on fully connected network (FCN) samples transfer and multi-classifier fusion is proposed in this study. The distribution similarity-based source domain selection method seeks source domains with marginal distributions that are similar to the target domain. Then, an FCN-based samples transfer method is proposed for transferring suitable samples from similar source domains to the target domain with similar conditional distributions. Additionally, a method for MI recognition is proposed that is based on the fusion of multiple classifiers. The classifiers are trained on labeled samples from the target domain as well as samples transferred from similar source domains. The new sample in the target domain can be identified using the weight fusion of the results of these classifiers. To evaluate the proposed method's effectiveness, four types of MI from the brain–computer interface competition IV dataset 2A were used to evaluate the recognition ability, and the results confirmed excellent recognition and generalization performance compared to commonly used methods nowadays. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 09410643 |
| DOI: | 10.1007/s00521-022-07748-7 |
