论文部分内容阅读
提出一种模式识别算法——双层支持量机算法,用来提高表面肌电识别精度.该算法融合集成学习中元学习的并行方法和叠加法的递进思想,把基本SVM分类器并行分布在第1层,第1层的预测结果作为第2层的输入,由第2层再进行分类识别,从而通过多层分类器组合来融合多源特征.以手臂表面肌电数据集为测试数据,采用文中的双层支持向量机,各肌肉的肌电信号分别输入基支持向量机,组合器融合各肌肉电信号特征,集成识别前臂肌肉群的肌电信号,从而实现运动意图的精确识别.实验结果显示,在预测精度上,此算法优于单个SVM分类器.在预测性能上(识别精度、耗时、鲁棒性),此算法优于随机森林和旋转森林等集成分类器.
This paper presents a pattern recognition algorithm called double support machine algorithm, which is used to improve the surface EMG recognition accuracy. The algorithm combines the parallel method of meta-learning in integrated learning and the progressive idea of superposition method. The basic SVM classifier is distributed in parallel In the first layer, the first layer of the prediction results as the input of the second layer, and then from the second layer of classification and identification, through the multi-classifier combination to fusion multi-source features.Using the arm surface EMG data set as the test data , Using the double support vector machine in this paper, the EMG signals of each muscle are respectively input to the SVM. The combiner integrates the EMG signals and integrates the EMG signals of the forearm muscle groups so as to realize the accurate recognition of the motion intent. Experimental results show that the proposed algorithm is superior to single SVM classifier in predictive accuracy, which is superior to integrated classifiers such as random forest and rotating forest in predictive performance (recognition accuracy, time-consuming, robustness).