Scalable Algorithms for Adaptive Statistical Designs

We present a scalable, high-performance solution to multidimensional recurrences that arise in adaptive statistical designs. Adaptive designs are an important class of learning algorithms for a stochastic environment, and we focus on the problem of optimally assigning patients to treatments in clini...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:ACM/IEEE SC 2000 Conference (SC'00) s. 6
Hlavní autoři: Oehmke, R., Hardwick, J., Stout, Q.F.
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 2000
Témata:
Algorithm design and analysis
bandit models
Clinical trials
computational learning theory
Cost function
Design optimization
dynamic domain decomposition
Dynamic programming
experimental algorithms
load balancing
Medical treatment
memory-intensive computing
message-passing
Multidimensional systems
Performance analysis
sequential analysis
Standards development
Stochastic processes
ISBN:9780780398023, 0780398025
ISSN:1063-9535
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:We present a scalable, high-performance solution to multidimensional recurrences that arise in adaptive statistical designs. Adaptive designs are an important class of learning algorithms for a stochastic environment, and we focus on the problem of optimally assigning patients to treatments in clinical trials. While adaptive designs have significant ethical and cost advantages, they are rarely utilized because of the complexity of optimizing and analyzing them. Computational challenges include massive memory requirements, few calculations per memory access, and multiply-nested loops with dynamic indices. We analyze the effects of various parallelization options, and while standard approaches do not work well, with effort an efficient, highly scalable program can be developed. This allows us to solve problems thousands of times more complex than those solved previously, which helps make adaptive designs practical. Further, our work applies to many other problems involving neighbor recurrences, such as generalized string matching.
ISBN:9780780398023
0780398025
ISSN:1063-9535
DOI:10.1109/SC.2000.10026