Home > Research > Publications & Outputs > Active learning accelerated automatic heuristic...
View graph of relations

Active learning accelerated automatic heuristic construction for parallel program mapping

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review

Published
  • William Ogilvie
  • Pavlos Petoumenos
  • Zheng Wang
  • Hugh Leather
Close
Publication date2014
Host publicationPACT '14 Proceedings of the 23rd international conference on Parallel architectures and compilation
Place of PublicationNew York
PublisherACM
Pages481-482
Number of pages2
ISBN (print)9781450328098
<mark>Original language</mark>English
EventThe 23rd International Conference on Parallel Architectures and Compilation Techniques - , United Kingdom
Duration: 23/08/201429/08/2014

Conference

ConferenceThe 23rd International Conference on Parallel Architectures and Compilation Techniques
Country/TerritoryUnited Kingdom
Period23/08/1429/08/14

Conference

ConferenceThe 23rd International Conference on Parallel Architectures and Compilation Techniques
Country/TerritoryUnited Kingdom
Period23/08/1429/08/14

Abstract

Building effective optimization heuristics is a challenging task which often takes developers several months if not years to complete. Predictive modelling has recently emerged as a promising solution, automatically constructing heuristics from training data, however, obtaining this data can take months per platform. This is becoming an ever more critical problem as the pace of change in architecture increases. Indeed, if no solution is found we shall be left with out of date heuristics which cannot extract the best performance from modern machines.

In this work, we present a low-cost predictive modelling approach for automatic heuristic construction which significantly reduces this training overhead. Typically in supervised learning the training instances are randomly selected to evaluate regardless of how much useful information they carry, but this wastes effort on parts of the space that contribute little to the quality of the produced heuristic. Our approach, on the other hand, uses active learning to select and only focus on the most useful training examples and thus reduces the training overhead.

We demonstrate this technique by automatically creating a model to determine on which device to execute four parallel programs at differing problem dimensions for a representative CPU-GPU based system. Our methodology is remarkably simple and yet effective, making it a strong candidate for wide adoption. At high levels of classification accuracy the average learning speed-up is 3x, as compared to the state-of-the-art.