Scholarlink

As smartphone applications become increasingly integrated into daily life, ensuring their reliability in diverse contexts is crucial. This study investigates how various contextual factors—such as Wi-Fi connectivity, screen activity, and battery status—impact application behavior and device performance. Using the CM-SPAM algorithm from the SPMF library to preprocess context event data from Android smartphones, we apply machine learning models i.e., Support Vector Machine (SVM), Random Forest (RF), and Long Short-Term Memory (LSTM) to predict application usage and behavior. Our results show that one-minute time interval yields the best prediction accuracy, with RF achieving the highest performance (F1-score of 0.842), outperforming both SVM and LSTM. These findings emphasize the importance of context-aware event data for improving the reliability and accuracy of smartphone applications, suggesting that context-driven models like RF can significantly enhance app performance across varying conditions.

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

Context-Aware Event Sequence Prediction for Sustainable Mobile Computing Using LSTM: A Deep Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Pooja Goyal,
  • Caiden Chadwick,
  • Melody Jing,
  • Samuel Williams,
  • Renee Bryce

摘要

As smartphone applications become increasingly integrated into daily life, ensuring their reliability in diverse contexts is crucial. This study investigates how various contextual factors—such as Wi-Fi connectivity, screen activity, and battery status—impact application behavior and device performance. Using the CM-SPAM algorithm from the SPMF library to preprocess context event data from Android smartphones, we apply machine learning models i.e., Support Vector Machine (SVM), Random Forest (RF), and Long Short-Term Memory (LSTM) to predict application usage and behavior. Our results show that one-minute time interval yields the best prediction accuracy, with RF achieving the highest performance (F1-score of 0.842), outperforming both SVM and LSTM. These findings emphasize the importance of context-aware event data for improving the reliability and accuracy of smartphone applications, suggesting that context-driven models like RF can significantly enhance app performance across varying conditions.