针对传统集成学习方法忽略不同样本需使用不同模型权重的问题,提出一种基于类权重和最小化预测熵(class and entropy weights,CEW)的测试时集成方法.类权重为模型预测结果与验证集上各类概率对错分布的相似度,利用欧氏距离计算相识度;在最小化熵过程中,线性组合模型预测经过类权重模块加权后的输出,寻找最小预测熵对应的线性组合作为熵权重,提高集成模型预测能力.试验结果表明:在 4 个公开医学图像数据集上,CEW方法与最优单一模型相比,平均召回率提高0.23%～2.81%,准确率提高 0.5%～2.54%;与DS方法相比,CEW方法平均召回率最多提高 1.25%,准确率最多提高 1.1%.基于CEW的测试时集成方法能够在测试时(无标签情况下)动态调整模型权重,比同类方法的预测精度更高.
A test-time ensemble method based on class weights and prediction entropy minimization
To address the issue of traditional ensemble learning methods overlooking the necessity for different model weights for varied samples,a test-time ensemble approach based on class and entropy weights(CEW)was proposed.Class weights were deter-mined by the similarity between the model's predictive results and the distribution of correct and incorrect probabilities for each class on the validation set,calculated using Euclidean distance.During the entropy minimization process,the output from the linear com-bination of model predictions was weighted by the class weight module.The linear combination corresponding to the minimum pre-dictive entropy was identified as the entropy weight,enhancing the predictive capability of the ensemble model.Experimental results showed that on four public medical image datasets,compared to the optimal single model,the CEW method improved the average recall rate by 0.23%to 2.81%,and accuracy by 0.5%to 2.54%.Compared to the DS method,the CEW method improved the aver-age recall rate by up to 1.25%and accuracy by up to 1.1%.The test-time ensemble method based on CEW proved capable of dy-namically adjusting model weights during testing(in an unlabeled situation),achieving higher prediction accuracy than similar methods.
NETL
NSTL
万方数据
