1 / 7

7 slides about utility theory

7 slides about utility theory. Any more wouldn’t be worth the effort. Maximum expected utility. Utility = U(o) Probability of o: P(o) Expected utility: P(o)U(o)

Pat_Xavi
Download Presentation

7 slides about utility theory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. 7 slides about utility theory Any more wouldn’t be worth the effort

  2. Maximum expected utility • Utility = U(o) • Probability of o: P(o) • Expected utility: P(o)U(o) • “If agent maximized utility function that correctly reflects performance, it will achieve the highest possible performance on average” (Duh)

  3. Utility theory notation • A > B: A prefers B • A ~ B: A is indifferent to B • A >~ B: A either is indifferent or prefers B • A and B are atomic states or “lotteries” : an ordered set of probabilities and values. Might be just one thing (with probability 1).

  4. Reasonable assumptions • Transitive: A > B, B>C => A>C (avoids a different kind of sucker bet). • Orderable: (A>B) or (A<B) or (A~B) • Continuity: A>B>C => p [p,A; 1-p,C] ~B (indifferent between getting B for sure, and getting either A or C).

  5. More assumptions • Substitutability: If A~B, then you can substitute B for A (and vice versa) in other lotteries. • Montonicity: A > B implies perfer lotteries that yield A with more probability that B. • Decomposability: Can reduce lotteries to simpler lotteries using probabilities.

  6. What kinds of measures? • Money is always good. • Micromorts (1/106 chance of dying)=$30. • QALY (a year of life with no infirmities)

  7. The value of information • Gathering information takes effort, • Having information must have a value • Really, not much different from any other collected resource.

More Related