机器学习基石 - Noise and Error

机器学习基石上 (Machine Learning Foundations)—Mathematical Foundations
Hsuan-Tien Lin, 林轩田,副教授 (Associate Professor),资讯工程学系 (Computer Science and Information Engineering)

Noise and Probabilistic Target

确定的改为概率,保证同一分布下

练习题

机器学习基石 - Noise and Error_第1张图片

Error Measure

  • out of sample
  • pointwise
  • classification
  • 0/1 error often for classification; squared error often for regression
  • extended VC theory/philosophy works for most H H and err

Algorithmic Error Measure

  • err is application/user-dependent
  • 学习流程图
    机器学习基石 - Noise and Error_第2张图片

Weighted Classification

  • 不同的情形有不同的权重
  • Weighted Pocket Algorithm

你可能感兴趣的:(机器学习基石)