talktiezzi

This shows you the differences between two versions of the page.

Both sides previous revision Previous revision Next revision | Previous revision | ||

talktiezzi [2014/07/25 16:15] imt |
talktiezzi [2014/07/25 17:26] (current) unige |
||
---|---|---|---|

Line 7: | Line 7: | ||

**Abstract:** Large optimization problems tend to be overly complex to solve and often a globally optimal solution may be impossible to find. For this reason specific strategies are needed to solve them. We propose an approach for the coordination of declarative knowledge, that is the exact specification of the complete optimization problem, and procedural knowledge, that is the specific knowledge about subproblems and their, possibly approximated, resolution strategies. We consider Soft Constraint Satisfaction Problems (SCSPs) and we introduce a formalism, similar to a process calculus, for their specification. Cost functions are associated to terms and form a model of such specification, where operators are interpreted as optimization steps. We apply our approach to a problem studied in the ASCENS e-mobility case study, for which we provide a model in terms of cost functions. The procedural part concerns heuristic choices about which dynamic programming strategy should be employed and how different ad-hoc approximation heuristics could be applied. | **Abstract:** Large optimization problems tend to be overly complex to solve and often a globally optimal solution may be impossible to find. For this reason specific strategies are needed to solve them. We propose an approach for the coordination of declarative knowledge, that is the exact specification of the complete optimization problem, and procedural knowledge, that is the specific knowledge about subproblems and their, possibly approximated, resolution strategies. We consider Soft Constraint Satisfaction Problems (SCSPs) and we introduce a formalism, similar to a process calculus, for their specification. Cost functions are associated to terms and form a model of such specification, where operators are interpreted as optimization steps. We apply our approach to a problem studied in the ASCENS e-mobility case study, for which we provide a model in terms of cost functions. The procedural part concerns heuristic choices about which dynamic programming strategy should be employed and how different ad-hoc approximation heuristics could be applied. | ||

- | **Slides:** {{:cina_genova_tiezzi.pdf|pdf file}} | + | {{:cina_genova_tiezzi.pdf|slides}} |

talktiezzi.1406297709.txt.gz ยท Last modified: 2014/07/25 16:15 by imt