Skip to content

TonyNgcn/SpeechModel

Repository files navigation

SpeechModel

我们总共训练了两个模型,SVM 和 NN。 SVM模型基于scikit-learn 编写,但是在测试集上的准确率大概在70%左右。 SVM模型有两个超参数,经过多次取值,发现γ的取值稳定在7.464263932294464

NN模型基于tensorflow编写,最后一次的训练数据如下: Saving dict for global step 160000: accuracy = 0.79023886, accuracy_baseline = 0.5109034, auc = 0.8691241, auc_precision_recall = 0.87003034, average_loss = 0.45028788, global_step = 160000, label/mean = 0.5109034, loss = 22.237293, precision = 0.7929293, prediction/mean = 0.49435177, recall = 0.79776424

在测试集上的准确率是79%

综上,我们提交的作业中,result.csv 是由第二种模型得出的 我们把作业一的资料传到了github上,地址为: Github: https://github.com/TonyNgcn/SpeechModel

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published