2011-01-27
Combining Strategies Efficiently: High-Quality Decisions from Conflicting Advice
Publication
Publication
We base countless practical decisions and actions on advice from experts every day. Unfortunately, experts don’t always agree with each other. Wouter Koolen designed new computer programs that automatically learn which advice should be followed. These programs also know exactly how to switch between the advice of different experts over time. This research is part of online learning, an emerging discipline at the intersection of computer science, information theory, and statistics. Koolen describes applications ranging from stock market investing, probability forecasting, and data compression to route planning. In each of these cases, he mathematically proves that his computer programs are just as successful as a hypothetical decision-maker with complete knowledge of the future, minus the low learning costs.
| Additional Metadata | |
|---|---|
| , | |
| P.M.B. Vitányi (Paul) , P.D. Grünwald (Peter) | |
| Universiteit van Amsterdam | |
| hdl.handle.net/11245/1.329428 | |
| ILLS Dissertation Series ; 2011-01 | |
| Combinatorial Algorithms in Bio-informatics | |
| Organisation | Algorithms and Complexity |
|
Koolen-Wijkstra, W. (2011, January 27). Combining Strategies Efficiently: High-Quality Decisions from Conflicting Advice. ILLC Dissertation Series. Retrieved from http://hdl.handle.net/11245/1.329428 |
|
