Online AlgorithmsNot offered this year
Offered discontinued, last taught:
Decisions must often be made before the entire data is available. Online algorithms solve problems in which commitments must be made as the data is arriving. Choosing which items to evict from a cache before knowing future requests, which advertisers to consider for displaying ads alongside the result of a search, or which most representative data to store when computing statistics about a huge stream of information. We will discuss worst-case model, which hinges against the worst possible future data, and some stochastic and game-theoretic models.