Couldn't this be a dangerous approach? If the error threshold changes because you're using a different metric or something (I don't know too much about the domain area), then your code is now completely broken.
This comment was marked helpful 0 times.
yihuaz
We set the error threshold before we start the iterations. The point here is that different algorithm might give us slightly "different solution", but as long as the solution converges to within error threshold, the algorithm is still valid.
Couldn't this be a dangerous approach? If the error threshold changes because you're using a different metric or something (I don't know too much about the domain area), then your code is now completely broken.
This comment was marked helpful 0 times.
We set the error threshold before we start the iterations. The point here is that different algorithm might give us slightly "different solution", but as long as the solution converges to within error threshold, the algorithm is still valid.
This comment was marked helpful 0 times.