Feature Selection Method for Machine Learning Algorithms

   A feature selection method for training various machine learning algorithms is described. It is based on well-known feature selection methods and can be  used for data processing to solve the classification problem using machine learning algorithms. The method consists of several stages: calculation of the score of each feature using the existing feature shuffling method based on several scoring parameters of the machine learning model, processing the collected data array for division into two classes (relevant and irrelevant features) using the K-means clustering algorithm, removal of irrelevant features from the general dataset to train the SVM classifier, and assessment of the classification accuracy of the algorithm. A uniqueness of the method lies in the use of several scoring parameters at once to improve the accuracy and flexibility of the model, as well as in the use of an ensemble of machine learning algorithms to select the best features. A number of experiments have been carried out to determine the effectiveness of the method. As a set of input data for the classifier, electromyographic muscle activity signal readings have been collected by a specialized sensor, where each data set corresponds to a special gesture (class). During the processing, a number of features have been extracted from the signal and selected using the developed method to compile the input dataset for further training of the SVM classifier. The trained model has been used to interpret gestures into control commands for the robotic device in real time. The application of the technique provides a higher accuracy of gesture recognition compared to methods that involve only one scoring parameter of the machine learning model for feature selection.

1. Quitadamo L. R., Cavrini F., Sbernini L., Riillo F., Bianchi L., Seri S., Saggio G. Support vector machines to detect physiological patterns for EEG and EMG-based human–computer interaction: a review // Journal of neural engineering, 2017. V. 14. № 1. P. 011001.

2. Budko R. Yu., Chernov N. N., Budko, A. Yu. Raspoznavanie myshechnyh usilij po signalu licevoj elektromiogrammy v rezhime real’nogo vremeni [Recognition of muscle efforts by the signal of the facial electromyogram in real time]. Nauchnyj vestnik Novosibirskogo gosudarstvennogo tekhnicheskogo universiteta, 2018, no. 2, pp. 59–74. (in Russian).

3. Semendarov A. V. Vybor struktury i giperparametrov SVM dlya effektivnogo resheniya zadach klassifikacii signalov elektromiografii [Selection of structure and hyper-parameters of SVM for efficient solution of the tasks of classification of electromyography signals]. Nauchnyj vzglyad v budushchee, 2019, vol. 1, no.14, pp. 23–33. (in Russian).

4. Khokhar Z. O., Xiao Z. G., Menon C. Surface EMG pattern recognition for real-time control of a wrist exoskeleton. // Biomedical engineering online, 2010. V. 9. № 1. P. 1–17. doi: 10.1186/1475-925X-9-41

5. Brownlee J. How to choose a feature selection method for machine learning // Machine Learning Mastery, 2019. V. 10.

6. Soledad G. Alternative Feature Selection Methods in Machine Learning. Available at: https://trainindata.medium.com/alternative-feature-selection-methods-in-machine-learning-7ed8142d594 (accessed 04. 04. 2022)

7. Pedregosa F., Varoquaux G., Gramfort A., Michel V., Thirion B., Grisel O., Blondel M., Prettenhofer P., Weiss R., Dubourg V., Vanderplas J., Passos A., Cournapeau D., Brucher M., Perrot M., Duchesnay E. Scikit-learn: Machine learning in Python // Journal of machine Learning research, 2011. V. 12. P. 2825–2830.

8. Voznenko T. I., Gridnev A. A., Kudryavtsev K. Y., Chepin E. V. The Decomposition Method of Multi-channel Control System Based on Extended BCI for a Robotic Wheelchair // Advances in Intelligent Systems and Computing, 2020. V. 948. P. 562–567. doi: 10.1007/978-3-030-25719-4_73

9. Voznenko T. I., Gridnev A. A., Chepin E. V., Kudryavtsev K. Y. The command interpretation in decomposition method of multi-channel control for a robotic device // Procedia Computer Science, 2020. V. 169. P. 152–157. doi: 10.1016/j.procs.2020.02.127

10. Voznenko T. I., Gridnev A. A., Chepin E. V., Kudryavtsev K. Y. Comparison Between Coordinated Control and Interpretation Methods for Multi-channel Control of a Mobile Robotic Device // Advances in Intelligent Systems and Computing, 2021. V. 1310. P. 558–564. doi: 10.1007/978-3-030-65596-9_68

11. Kapur A., Kapur S., Maes P. AlterEgo: A personalized wearable silent speech interface // In 23rd International conference on intelligent user interfaces, 2018. P. 43–53. doi: 10.1145/3172944.3172977