Scholarlink

<p>With the ever-increasing population and the parallel rise of affordable automobiles, there has been a boom in the number of vehicles on the road in the last decade. However, this growth has not been matched with adequate safety measures, resulting in many road accidents that are still prevalent today. Researchers have focused on ascertaining classification severity using existing machine and deep learning approaches but failed to synchronize both computational time and system performance simultaneously. Hence, there is a need for an intelligent system that can efficiently manage this synchronization problem. In this vein, this research work proposed an efficient and reliable approach termed ‘DELM-SVM’ where Deep ELM (DELM) non-linearly maps the input features to a higher dimensional feature space and linearly separates these features by using the minimum norm with output weight property. Further, the Support Vector Machine (SVM) takes these linearly separated features as input and applies its marginal property with maximum separation capability for efficient classification. The experimental findings reveal that the proposed DELM-SVM approach surpasses current state-of-the-art approaches in two key aspects: firstly, it significantly reduces training time, and secondly, it delivers better performance.</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

Road crash severity classification: a novel integrated approach

  • 期刊论文

  • 发表日期：

    2026年5月21日

查看全文
  • Rajendra Kumar Roul,
  • Samridhi Mangla

摘要

With the ever-increasing population and the parallel rise of affordable automobiles, there has been a boom in the number of vehicles on the road in the last decade. However, this growth has not been matched with adequate safety measures, resulting in many road accidents that are still prevalent today. Researchers have focused on ascertaining classification severity using existing machine and deep learning approaches but failed to synchronize both computational time and system performance simultaneously. Hence, there is a need for an intelligent system that can efficiently manage this synchronization problem. In this vein, this research work proposed an efficient and reliable approach termed ‘DELM-SVM’ where Deep ELM (DELM) non-linearly maps the input features to a higher dimensional feature space and linearly separates these features by using the minimum norm with output weight property. Further, the Support Vector Machine (SVM) takes these linearly separated features as input and applies its marginal property with maximum separation capability for efficient classification. The experimental findings reveal that the proposed DELM-SVM approach surpasses current state-of-the-art approaches in two key aspects: firstly, it significantly reduces training time, and secondly, it delivers better performance.