Conclusion of Lecture 20

• Machine learning algorithms can be used to learn virtually everything.
• They depend on training data, so the system gets better with more data.
• The more you know about the data, the easier it is to build a system, and the better the result.
• Many techniques are applicable to each problem.
• No technique is best for every problem.
• The reading for this week was flexible. You needed to pick a paper on Machine Learning.
• There is no reading for next week.