Scholarlink

In this paper, we propose a novel random oblique stumps algorithm (ROS) using the graphics processing unit (GPU) to perform the classification task of the ImageNet challenge. The ROS-GPU algorithm bases on our proposed multi-class linear discriminant analysis (MC-LDA) on GPU, with the One-Versus-All (OVA) multi-class strategy and the under-sampling technique to train in the parallel manner ensemble ROS classifiers. The empirical test results on the ImageNet dataset show that our ROS-GPU algorithm with the GPU (Gigabyte GeForce RTX 3060 12 GB GDDR6, 3584 CUDA cores) is faster and more accurate than the state-of-the-art parallel SVM algorithms and the parallel covertree-kNN run on a PC (Intel(R) Core i7-4790 CPU, 3.6 GHz, 4 cores, 32 GB RAM).

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

ImageNet Classification Using the Random Oblique Stumps Algorithm on the Graphics Processing Unit

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Minh-Trung Nguyen,
  • Truong-Thanh Ma,
  • Thanh-Nghi Do

摘要

In this paper, we propose a novel random oblique stumps algorithm (ROS) using the graphics processing unit (GPU) to perform the classification task of the ImageNet challenge. The ROS-GPU algorithm bases on our proposed multi-class linear discriminant analysis (MC-LDA) on GPU, with the One-Versus-All (OVA) multi-class strategy and the under-sampling technique to train in the parallel manner ensemble ROS classifiers. The empirical test results on the ImageNet dataset show that our ROS-GPU algorithm with the GPU (Gigabyte GeForce RTX 3060 12 GB GDDR6, 3584 CUDA cores) is faster and more accurate than the state-of-the-art parallel SVM algorithms and the parallel covertree-kNN run on a PC (Intel(R) Core i7-4790 CPU, 3.6 GHz, 4 cores, 32 GB RAM).