deletions | additions       diff --git a/chapter_Discussion_label_chap_discussion__.tex b/chapter_Discussion_label_chap_discussion__.tex  index c6da039..704f58f 100644  --- a/chapter_Discussion_label_chap_discussion__.tex  +++ b/chapter_Discussion_label_chap_discussion__.tex 
...
In this chapter, the results of the experiments are discussed with regards to accuracy, frequency, and future improvements. Afterward, in the General Discussion, the research questions are addressed and discussed.  The comparison of the amount of samples with the average cosine  similarity between sub-sampling and full sampling  (Experiment~\ref{sec:numtextons}) has shown that only a small part of the maximum amount of samples suffices. In fact, $\frac{400}{640 \times 480} = 0.13\,\%$  suffices to achieve similarities larger than $99\,\%$.In  fact, $\frac{400}{640 \times 480} = 0.13\,\%$.  This set the stage for large speed-ups during live operation. Additionally, it allow the technique allows  for further speed-ups depending on the processing power of the used  CPU. platform at hand.