deletions | additions       diff --git a/chapter_Analysis_label_chap_analysis__.tex b/chapter_Analysis_label_chap_analysis__.tex  index 0fd775a..fbad894 100644  --- a/chapter_Analysis_label_chap_analysis__.tex  +++ b/chapter_Analysis_label_chap_analysis__.tex 
...
similarity was determined as well as the standard deviation of the  cosine similarity was measured. Comparing the cosine similarity  between the histograms has the advantage that the number of samples  can be determined independent of a specific task.\section{Analysis -- Setting the Baseline for kNN and determining k}  \label{sec:numtextons}  In a standard setting, the training error $\epsilon_t$ of a  $k$=1-nearest neighbor algorithm is $\epsilon_t = 0$ because the  nearest neighbor of the sample will be the sample itself. However, in  this scenario, we deal with random sampling such that each image will  be represented by a slightly different histogram each time the  histogram is extracted.