on March 26, 2015 at 2:00 pm
Polaris Colloquium: Cancelled
Why test manually if one can generate tests? Recent advances in genetic algorithms automatically explore the behavior even of internet-scale applications by systematically evolving a population of inputs (keystrokes, clicks, settings) towards a testing goal such as coverage. To check the results of these tests, I present recent research that automatically mines patterns of normal behavior from a multitude of instances - for instance, by comparing web site behavior across multiple browsers, or comparing app behavior againt similar apps from app stores. These techniques easily scale to industrial settings, and companies like Google or our Testfabrik startup are rapidly adopting them for full-fledged testing systems.
More...Auditorium IRCICA, parc scientifique de la Haute Borne à Villeneuve d'Ascq