Currently, there is only one worker and this worker makes use of Dynamic Time Warping (DTW). The execution time for classification increases linear with the number of already classified handwritings.
About 4000 - 5000 was too much for this technique / my web server.
An alternative that might be faster with classification (and slower with learning) and eventually lead to better classifications are neural nets. So I should write a classifier that makes use of neural nets.
Currently, there is only one worker and this worker makes use of Dynamic Time Warping (DTW). The execution time for classification increases linear with the number of already classified handwritings.
About 4000 - 5000 was too much for this technique / my web server.
An alternative that might be faster with classification (and slower with learning) and eventually lead to better classifications are neural nets. So I should write a classifier that makes use of neural nets.