https://www.youtube.com/watch?v=UzxYlbK2c7E
Four parts of the class:
- Supervise learning: providing the computer existing datasets that have the right answers.
- Regression (continuous data point) or classification (discrete data) are example of supervise learning
- Learning theory: understand how and why learning algorithm work (how to prove that algorithm that reads zip code works); what algorithm can approximate function, what size for the learning data we need.
- Unsupervised learning: being given a dataset and ask to find interesting structure (vs. giving the right answer)
- Clustering is one example
- Reinforcement learning: asked to make a sequence of decision over time;
- reinforce good behavior vs bad behavior: example flying helicopter: good behavior doesn't crash.
