Merge pull request #87 from CogSciUOS/ToBeMaster

correct spelling

documentation/report/Chapters/Classification.tex

@@ -44,7 +44,7 @@ \subsubsection{Prediction based on feature engineering}
 
 An extensive search in hyperparameter space is practicable for simple \acrshortpl{mlp}. This is because of them having few parameters only and because training on sparse representations limits the number of neurons in the networks. Taken together this results in fast training that allows for many experiments. If the learning task is simple enough to be accomplished by \acrshortpl{mlp} (e.g.\ finding combinations of partial angles that correspond to the impression of curvature), one may hence speculate that underfitting can rather be explained by incongruencies or missing information in the labels than a result of issues in network design and training parameters. 
 
-This classical machine learning approach which relies on feature engineering is applied to predict features based on color and partial angles of asparagus spears because of these benefits.\footnote{~See \url{https://github.com/CogSciUOS/asparagus/tree/FinalProject/classification/supervised/mlps/\_and\_feature\_engineering}}
+This classical machine learning approach which relies on feature engineering is applied to predict features based on color and partial angles of asparagus spears because of these benefits.\footnote{~See \url{https://github.com/CogSciUOS/asparagus/tree/FinalProject/classification/supervised/mlps\_and\_feature\_engineering}}
 
 \bigskip
 \textbf{Violet and rust prediction based on color histograms}