Edinburgh Research Explorer

End-to-end Deep Learning of Optimization Heuristics

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Related Edinburgh Organisations

Open Access permissions

Open

Documents

http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8091247
Original languageEnglish
Title of host publication2017 26th International Conference on Parallel Architectures and Compilation Techniques (PACT)
Number of pages14
ISBN (Electronic)978-1-5090-6764-0
DOIs
StatePublished - 2 Nov 2017
Event26th International Conference on Parallel Architectures and Compilation Techniques - Portland, United States
Duration: 9 Sep 201713 Sep 2017
https://parasol.tamu.edu/pact17/

Conference

Conference26th International Conference on Parallel Architectures and Compilation Techniques
Abbreviated titlePACT 2017
CountryUnited States
CityPortland
Period9/09/1713/09/17
Internet address

Abstract

Accurate automatic optimization heuristics are necessary for dealing with the complexity and diversity of modern hardware and software. Machine learning is a proven technique for learning such heuristics, but its success is bound by the quality of the features used. These features must be hand crafted by developers through a combination of expert domain knowledge and trial and error. This makes the quality of the final model directly dependent on the skill and available time of the system architect.

Our work introduces a better way for building heuristics. We develop a deep neural network that learns heuristics over raw code, entirely without using code features. The neural network simultaneously constructs appropriate representations of the code and learns how best to optimize, removing the need for manual feature creation. Further, we show that our neural nets can transfer learning from one optimization problem to another, improving the accuracy of new models, without the help of human experts.

We compare the effectiveness of our automatically generated heuristics against ones with features hand-picked by experts. We examine two challenging tasks: predicting optimal mapping for heterogeneous parallelism and GPU thread coarsening factors. In 89% of the cases, the quality of our fully automatic heuristics matches or surpasses that of state-of-the-art predictive models using hand-crafted features, providing on average 14% and 12% more performance with no human effort expended designing features.

Event

26th International Conference on Parallel Architectures and Compilation Techniques

9/09/1713/09/17

Portland, United States

Event: Conference

Download statistics

No data available

ID: 37747686