Scholarlink

<p>In this paper, a deep learning driven multidimensional evaluation framework is proposed to evaluate the performance of Youth Training System (YTS). In order to reduce the subjectivity of manual scoring and static weight, the framework organizes resources organization, training execution, competitive transformation, support guarantee, and development output into a unified system. Multiple sources collection, anomaly filtering, robust standardization, rolling-window reconstruction, hierarchical coding, cross-dimension feature fusion, and joint score-output modeling have been specified. Experiments were carried out on a total of 1, 536 effective sample windows from 32 Youth Training Units, including repeated runs, tuned baseline, ablative, robust, cross-sample transfer, and deployment. Across the assessed data set and experimental settings, the proposed model obtained a total accuracy of 91.84%, a F1 score of 0.903, a MAE of 0.146, and a RMSE of 0.221. Dimension-level results indicate that the performance of training and the efficiency of resource organization are more stable, but the competition transformation is still more difficult to predict. The explanation analysis further maps the output of the model into an interpretable management factor, which can be used to diagnose, classify, and alert the YTS performance management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design of a deep learning-driven multidimensional evaluation model for youth training system performance assessment

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月2日

查看全文
  • Bingmei Yu,
  • Wei Li,
  • Chuanzhi Zhou

摘要

In this paper, a deep learning driven multidimensional evaluation framework is proposed to evaluate the performance of Youth Training System (YTS). In order to reduce the subjectivity of manual scoring and static weight, the framework organizes resources organization, training execution, competitive transformation, support guarantee, and development output into a unified system. Multiple sources collection, anomaly filtering, robust standardization, rolling-window reconstruction, hierarchical coding, cross-dimension feature fusion, and joint score-output modeling have been specified. Experiments were carried out on a total of 1, 536 effective sample windows from 32 Youth Training Units, including repeated runs, tuned baseline, ablative, robust, cross-sample transfer, and deployment. Across the assessed data set and experimental settings, the proposed model obtained a total accuracy of 91.84%, a F1 score of 0.903, a MAE of 0.146, and a RMSE of 0.221. Dimension-level results indicate that the performance of training and the efficiency of resource organization are more stable, but the competition transformation is still more difficult to predict. The explanation analysis further maps the output of the model into an interpretable management factor, which can be used to diagnose, classify, and alert the YTS performance management.