Skip to Main content Skip to Navigation
Journal articles

Learning System for Defactorization Factor Classification of Factorized Data Dependence Graph

Abstract : In the presence of a concrete problem of multi-objective optimization, we are confronting with the principal difficulty to choice a method producing the optimal solutions. This choice implies the knowledge and the expertise of the user. In this framework, we are interested to the flow of design based on methodology AAA (Adequacy Algorithm Architecture). The extension of this methodology to the circuits allows the exploitation of potential parallelism onto components. It aims to obtaining a real time implementation witch respect the temporal constraint of the application while minimizing the resources. Then, from an algorithm specified with a data flow graph, this exploration of parallelism is NP-complete problem. In this work, we propose a new solution to perform a multi-objective exploration by integrating an SVM (Support Vector Machine) training aptitude to an agent. We validate our model by a simulation based on the greedy heuristic results of SynDEX-IC (Synchronized Distributed Executive for Integrated circuit) [1] examples.
Complete list of metadatas

https://hal-upec-upem.archives-ouvertes.fr/hal-00826253
Contributor : Thierry Grandpierre <>
Submitted on : Monday, May 27, 2013 - 11:20:21 AM
Last modification on : Wednesday, February 26, 2020 - 7:06:17 PM

Identifiers

  • HAL Id : hal-00826253, version 1

Citation

Rachida Saouli, Mohamed Akil, Thierry Grandpierre. Learning System for Defactorization Factor Classification of Factorized Data Dependence Graph. IJACT : International Journal of Advancements in Computing Technology, 2011, 3 (4), pp.1-13. ⟨hal-00826253⟩

Share

Metrics

Record views

386