Optimizing Embedded Applications

The course aims at providing the basics for designing optimal embedded applications starting from a given problem. The course, configured to stimulate the interaction with the students, will address the following methodological aspects - Problem complexity and complexity reduction (deterministic vs probabilistic approaches for problem solving; Randomized algorithms) - Approximate computing (sources of approximation, Probably approximately correct computation) - Optimization methods for embedded applications (gradient-based optimization, evolutionary-based optimization, learning mechanisms) - Application porting to low precision hardware platforms (robustness analysis in the small; robustness analysis in the large; accuracy loss estimation) - Performance and quality assessment of the solution (Crossvalidation, bootstrap, bags of little bootstraps).

REFERENCES

• C.Alippi, Intelligence for Embedded Systems: a Methodological approach, Springer, 2014
• Technical papers and reference material provided by the professor