Up to this point, all experiments have shown the results of EUREKA selecting a single strategy, all other strategy results being fixed. In this experiment we allow EUREKA to select all strategy choices at once for a given problem and execute the parallel search with the recommended strategies. We then compare the results to each fixed strategy (the fixed strategy choice is averaged over all problem instances and all possible choices of other strategy decisions). A random set of 50 problems from the fifteen puzzle domain is selected and run on 64 processors of the nCUBE 2. Table 13 summarizes the speedup for each approach.
These results indicate that EUREKA can effectively make all strategy choices at once. The learned rules achieve better performance than that obtained by any one of these strategy choices. These rules also outperform any single fixed strategy choice averaged over all other parameter options.