Lecture1: Intro
Artificial Intelligence¶
Brief Hisroty¶
- Symbolic AI
- Neural AI
- Statistical AI
Early AI Programs¶
- Checkers Player: lookahead search
- Newell & Simon’s Logic Theorist: search tree and heuristics search
Birth of AI¶
Workshop at Dartmouth College
- Name "Artificial Intelligence" for the new field
First AI Winter¶
-
Underwhelming results
- Machine Translation
-
Problems
- Limited computation: search space grew exponentially, outpacing hardware
- Limited information: complexity of AI problems
Knowledge-based Systems¶
- Expert systems
- Encode prior knowledge to reduce computation
Deterministic rules could not handle the uncertainty of the real world.
Neural Networks¶
- CNN
- backpropagation id popluar
- Recognizing handwritten digits for USPS
- consider AI tools, not to make a "real human"
- AlexNet
- AlphaGo
- deep reinforcement learning, MCTS
- Bayesian Networks
- graphical models, reasoning under uncertainty
- Support Vector Machines
- Find the best by optimization algorithm!
AI Methods¶
- Search
- Symbolic AI
- Nerual AI
- Statistical AI
- Communicating, Perceiving, Acting
Two Views of AI¶
Two Views¶
- Intelligence Agent
- how can we create intelligence
- AI Tools
- how can we benefit society
AI Tools¶
AI + Science