A folk belief about genetic algorithms is that they outperform alternative search based algorithms (e.g., hill climbing) because of the crossover operator, which allows for powerful recombinations of partial solutions. In this talk, we analyze the conditions under which this belief holds and the conditions under which it does not. The former can be formalized as a family of "royal road" fitness functions.
You are here
On: "Automated oracle comparators for testing web applications", by Sprenkle, S., Pollock, L., Esquivel, H., Hazelwood, B., Ecott, S. (ISSRE 2007).
Recent trends in the field of computing have highlighted the emergence of a form of human computing known as Crowdsourcing and platforms that make it accessible in a relatively scalable manner. Consequently, the issue of automated testing can be revisited in light of the existence of these platforms. In particular, one can ask: is it possible to exploit crowdsourcing platforms for addressing outstanding issues in automated software testing ?
Sample Size vs. Bias in Defect Prediction