Scholarlink

It is argued that machine learning methods are inductive methods and therefore rely on inductive assumptions. These assumptions are made explicit in two different justificatory strategies. Because inductive assumptions are not determined by the data machine Learning methods are trained on, they require human choices at the very outset. And these choices cannot be automated.

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

Inductive Assumptions in ML Always Required, Often Ignored

  • 章节

  • 发表日期：

    2025年

查看全文
  • Nico Formanék

摘要

It is argued that machine learning methods are inductive methods and therefore rely on inductive assumptions. These assumptions are made explicit in two different justificatory strategies. Because inductive assumptions are not determined by the data machine Learning methods are trained on, they require human choices at the very outset. And these choices cannot be automated.