From these preliminary experimental results, we see that the recognition performance can be improved with the collection of single condition models, when a single-condition model matches the run-time condition very well.
The first among these improvements was a technique suggested by Valiant [ 23], who showed that the CFG Parsing problem on a sentence with n words can be solved in a running time which matches the running time of a Boolean Matrix Multiplication of two n × n matrices.
The running time of the adapted algorithm is different from that of Valiant's algorithm, and matches the running time of a Max-Plus Multiplication of two n × n matrices.
It is stronger than the bounds that follow from general black-box complexity theory and is matched by the run time of many evolutionary algorithms.
Note that for a failed match, the run time can only be smaller, and therefore in this case GIMME has a worst-case run time in O(n).
We observed that outlier removal has a positive impact on matching run-time and storage requirements, while there are no significant differences in the localization error within the methods.
Requiring these sets of hits to be fully consistent with the CM would re-introduce the complexity that was eliminated in the unbranched match stage, bringing the running time back up to O(N).
The new MCM algorithms we developed allow our HSA algorithm to match, and even improve, the running times of the previous algorithms of [ 26, 27] for less general variants of the HSA problem.
The results of this comparison are listed in Table 5 and the running times of both programs, for matching and layouting, are shown in Table 6.
Only using advanced hardware (a total of 12 CPUs of 3 GHz each) and implementing extensive debugging allowed the reduction of the running time for a history match to nearly 75 h.
