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

Going Neural

  • Yannis Haralambous

摘要

A deep neural network can have millions of parameters. Where do they get their values from? Here is the trick: it starts with random values, calculates a first result starting from an observation (called propagation), and compares the result obtained with the expected result.