next up previous contents
Next: 7 Summation Up: 6 Synthesis Previous: 6.4 Time issues

6.5 Optimisations

There are many possible ways to optimise the learning algorithm. These include techniques such as forward and backward selection ([JKP94]) as well as other space-searching techniques. However, the application of these is non trivial and usually takes considerable time.

The conclusion should not therefore be drawn that these results are even near-optimal.



waleed@cse.unsw.edu.au