README.md

GeneralAI

一直对 AI 情有独钟，学习算法半年有余，感觉各方面知识过于庞杂：机器学习、深度学习、强化学习、GAN、传统算法……数不胜数，故决定构建自己的 AI 算法体系。

恰好看到 MIT 的人工智能课程，看了课程介绍觉得欲罢不能，决定就从这里开始。

AI 简史

• 1842: Lady Lovelace, First programmer
• 1950: Turing, Turing Test Paper
• 1960: Marvin Minsky, 《Steps toward Aitificial Intelligence》
• 1970-2000:
  • early DAWN age
    • James Slagle, a program did symbolic integration（积分程序）
    • Another: Eliza
    • Another: analogy
  • Late dawn
    • vision
    • learning
    • expert systems: Bizs, Deep Blue
  • BULLDOZER time

What Is AI?

• THINKING, PERCEPTION, ACTION: LOOPS TOGETHER
• MODELS TARGETED AT
• REPRESENTATIONS THAT SUPPORT
• CONSTRAINTS EXPOSED BY
• ALGORITHMS ENABLED BY

GENERATE AND TEST

• Generating some possible solutions
• feeding them into a box that test them
• out the other side comes mostly failures, but every once in a while we get something that succeeds and pleases us.

Rumpelstiltskin principal

once you can name something, you get power over it.

Lauguage

Lauguage is at the center of thins because oit enables storytelling going up, and marshaling the resources of the perceptual apparatus that are lying down. 语言处在一切的中心，它向上让故事讲述成为可能，并向下统御感觉系统的各种资源。

Sources

• deepmind/learning-to-learn: Learning to Learn in TensorFlow
• An Explicitly Relational Neural Network Architecture