论文部分内容阅读
音乐流派是区分和描述不同音乐的一种标签,借助数学和计算机的方法将大量音乐自动分为不同流派是目前国内外研究的热点问题之一.支持向量机(SVM)由于其具有严格的数学理论基础而被广泛应用于音乐流派自动分类.然而,支持向量机的惩罚参数和核参数对其分类效果具有重要影响.以交叉验证正确率作为适应值,采用人工蜂群(ABC)算法优化支持向量的控制参数.在音乐流派自动分类的仿真实验中,经ABC算法优化后的支持向量机取得的平均预测正确率为80.8000%(最优预测正确率达83%),高出默认参数SVM 18.8个百分点.与粒子群优化算法及遗传算法相比,仿真实验结果同样显示了ABC算法的优越性.
Music genre is a kind of label that distinguishes and describes different music.It is one of the hot topics at home and abroad that a large number of music are automatically divided into different genres by means of mathematics and computer.Support Vector Machine (SVM) Theoretical basis and has been widely used in music genres automatic classification.However, the penalty parameters and kernel parameters of SVM have an important impact on the classification results.With the correctness of cross-validation as an adaptive value, using artificial bee colony (ABC) algorithm to optimize support Vector control parameters.In the simulation experiment of music genre automatic classification, the average prediction accuracy obtained by the support vector machine optimized by ABC algorithm is 80.8000% (the optimal prediction accuracy rate is 83%), higher than the default parameter SVM 18.8 Percentage points.Compared with PSO and GA, simulation results also show the superiority of ABC algorithm.