<caner AT candan DOT fr> (PGP1, PGP2)
<candan AT info DOT univ-angers DOT fr> (PGP)
Thanks for your visit.
Parallelization and programmation models, stochastic optimization algorithms, artificial intelligence, optimization, meta-heuristics, experimental analysis, careful programming approach.
Caner Candan is a student researcher in high performance computing, stochastic optimization algorithms, artificial intelligence, optimization and metaheuristics.
Here's the list of achievements that I made:
Parallelization module in EO
Implementation of a parallelization module in the framework EO “Evolving Objects” in C++, using OpenMP. Scalability issues in variation and replacement operators due mainly to memory allocations not under control. Works with larger population sizes. Experiments, on a 48-core machine, showed significant speedups with an increasing number of cores. Available at http://eodev.sf.net (Published at GECCO 2011)
Linear algebra framework
Conception and implementation of a linear algebra computation framework in C++ along with a layout abstraction for parallelization paradigms. Provides operators to compute dense and sparse matrices with generically designed scalar, complex, vector and matrix types. Available at https://github.com/canercandan/linear-algebra
Conception and implementation of the EDO framework, for the design of estimation of distribution algorithms. Used along with ParadisEO-MO to implement an original hybridation of multi-variate gaussian EDA with simulated annealing. Available at http://eodev.sf.net (Published at ROADEF 2011)
Implementation of a multi-objective parameter setting and behaviour extraction method, using ParadisEO framework. http://paradiseo.gforge.inria.fr
Here is the projects list I am managing, contributing, or just supporting:
You can download the CV in clicking here.