Accelerating Machine Learning Algorithms with TensorFlow Using Thread Mapping Policies
| Autor: | Danilo Carastan-Santos, Matheus W. Camargo, Philippe O. A. Navaux, Alexandre Carissimi, Matheus da Silva Serpa |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
020203 distributed computing
Multi-core processor Computer science business.industry Inference 010103 numerical & computational mathematics 02 engineering and technology Python (programming language) Machine learning computer.software_genre 01 natural sciences Execution time Work (electrical) 0202 electrical engineering electronic engineering information engineering Thread mapping Artificial intelligence 0101 mathematics business computer Algorithm computer.programming_language |
| Zdroj: | Communications in Computer and Information Science ISBN: 9783030680343 |
| DOI: | 10.1007/978-3-030-68035-0_5 |
| Popis: | Machine Learning (ML) algorithms are increasingly being used in various scientific and industrial problems, with the time of execution of these algorithms as an important concern. In this work, we explore mappings of threads in multi-core architectures and their impact on new ML algorithms running with Python and TensorFlow. Using smart thread mapping, we were able to reduce the execution time of both training and inference phases for up to 46% and 29%, respectively. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|