To address the problem of existing active learning algorithms that typically ignore sample feature representation infor-mation during sampling,an active learning model was proposed based on sample feature representation and model prediction.To alleviate the cold-start problem caused by active learning algorithms in the early stages of model training,a labeling set initializa-tion algorithm was proposed.Clustering methods were utilized to extract sample feature representation information and the sam-ple model prediction information was obtained through a classifier.The sample distribution of the initial labeling set was made as similar as possible to that of the original dataset.Experimental results demonstrate that the proposed active learning model out-performs multiple active learning baseline algorithms in classification accuracy,and the labeling set initialization algorithm effec-tively alleviates the cold-start problem.
关键词
主动学习/特征表达/模型预测/冷启动/聚类/图像分类/标注集初始化
Key words
active learning/feature representation/model prediction/coldstart/clustering/image classification/labeling set ini-tialization
维普
万方数据