changed backdoor attacks, changed thesis

Author: EvilWatermelon

masterthesis_tex/masterthesis_main.pdf

@@ -1076,3 +1076,68 @@ @inproceedings{DBLP:conf/nips/RenHGS15
   biburl    = {https://dblp.org/rec/conf/nips/RenHGS15.bib},
   bibsource = {dblp computer science bibliography, https://dblp.org}
 }
+
+@inproceedings{Krizhevsky2009LearningML,
+  title={Learning Multiple Layers of Features from Tiny Images},
+  author={Alex Krizhevsky},
+  year={2009}
+}
+
+@article{DBLP:journals/corr/ArjovskyCB17,
+  author    = {Mart{\'{\i}}n Arjovsky and
+               Soumith Chintala and
+               L{\'{e}}on Bottou},
+  title     = {Wasserstein {GAN}},
+  journal   = {CoRR},
+  volume    = {abs/1701.07875},
+  year      = {2017},
+  url       = {http://arxiv.org/abs/1701.07875},
+  eprinttype = {arXiv},
+  eprint    = {1701.07875},
+  timestamp = {Mon, 13 Aug 2018 16:48:27 +0200},
+  biburl    = {https://dblp.org/rec/journals/corr/ArjovskyCB17.bib},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}
+
+@inproceedings{DBLP:conf/nips/GulrajaniAADC17,
+  author    = {Ishaan Gulrajani and
+               Faruk Ahmed and
+               Mart{\'{\i}}n Arjovsky and
+               Vincent Dumoulin and
+               Aaron C. Courville},
+  editor    = {Isabelle Guyon and
+               Ulrike von Luxburg and
+               Samy Bengio and
+               Hanna M. Wallach and
+               Rob Fergus and
+               S. V. N. Vishwanathan and
+               Roman Garnett},
+  title     = {Improved Training of Wasserstein GANs},
+  booktitle = {Advances in Neural Information Processing Systems 30: Annual Conference
+               on Neural Information Processing Systems 2017, December 4-9, 2017,
+               Long Beach, CA, {USA}},
+  pages     = {5767--5777},
+  year      = {2017},
+  url       = {https://proceedings.neurips.cc/paper/2017/hash/892c3b1c6dccd52936e27cbd0ff683d6-Abstract.html},
+  timestamp = {Thu, 21 Jan 2021 15:15:21 +0100},
+  biburl    = {https://dblp.org/rec/conf/nips/GulrajaniAADC17.bib},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}
+
+@article{DBLP:journals/corr/GulrajaniAADC17,
+  author    = {Ishaan Gulrajani and
+               Faruk Ahmed and
+               Mart{\'{\i}}n Arjovsky and
+               Vincent Dumoulin and
+               Aaron C. Courville},
+  title     = {Improved Training of Wasserstein GANs},
+  journal   = {CoRR},
+  volume    = {abs/1704.00028},
+  year      = {2017},
+  url       = {http://arxiv.org/abs/1704.00028},
+  eprinttype = {arXiv},
+  eprint    = {1704.00028},
+  timestamp = {Mon, 13 Aug 2018 16:47:43 +0200},
+  biburl    = {https://dblp.org/rec/journals/corr/GulrajaniAADC17.bib},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}

masterthesis_tex/pictures/original_example.png

@@ -3,36 +3,44 @@ \section{Case Study functions}
 \label{sec:case_study_functions}
 
 \textbf{class\_num}
-This argument \\ \\
+This argument gets the key value from the labels which is an integer. \\ \\
 \textbf{train\_number}
-This argument \\ \\
+This argument gets the number of images from a labeled folder. \\ \\
 \textbf{train\_path}
-This argument \\ \\
+This argument gets the path where the local training data is stored. \\ \\
 \textbf{data\_dir}
-This argument \\ \\
+This argument gets the path where all local images are stored. \\ \\
 \textbf{image\_data}
-This argument \\ \\
+This argument gets an array with all images from a label. \\ \\
 \textbf{image\_labels}
-This argument \\ \\
+This argument gets the label which belongs to the corresponding images. \\ \\
 
 \begin{lstlisting}
   dataset_visualization(class_num, train_number)
 \end{lstlisting}
 
-\noindent In this function the number of images are visualized and sorted from the lowest to the highest number of images per label.
+\noindent With this function, the number of images are visualized and sorted from the lowest to the highest number of images per label.
 
 \begin{lstlisting}
   read_training_data(train_path, data_dir)
 \end{lstlisting}
 
+\noindent This function reads in the training data, calls the \textit{dataset\_visualization()} function and resize the images to $30x30$ pixels. The function is called by the \textit{preprocessing()} function.
+
 \begin{lstlisting}
   preprocessing(train_path, data_dir, image_data, image_labels)
 \end{lstlisting}
 
+\noindent After calling the \textit{read\_training\_data()} function, this function assign the \textbf{image\_date} and \textbf{image\_labels} arguments to shuffle the training data. Then the training data splits into training and validation data. The shape of the images must be reshaped for the SVM.
+
 \begin{lstlisting}
   model_training(train_path, data_dir, image_data, image_labels)
 \end{lstlisting}
 
+\noindent After calling the \textit{preprocessing()} function, this function calls a pipeline function with the \textit{SVC()} class and then fits the classifier.
+
 \begin{lstlisting}
   read_test_data(train_path, data_dir, image_data, image_labels)
 \end{lstlisting}
+
+\noindent After calling the \textit{model\_training()} function, this function reads in the test data, resizes the images to $30x30$ pixels and then reshape the images. The last step is the prediction with the test data.