AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK
Steganography is a method for hiding secret messages on media cover in the form of text, audio, images or videos, so that the message is not suspected by those who are not authorized to open the message. The technique to find out whether the cover media is a stego file or not is steganalysis. In...
Saved in:
Main Author: | |
---|---|
Format: | Final Project |
Language: | Indonesia |
Online Access: | https://digilib.itb.ac.id/gdl/view/39479 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Institut Teknologi Bandung |
Language: | Indonesia |
id |
id-itb.:39479 |
---|---|
spelling |
id-itb.:394792019-06-26T14:17:27ZAUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK Rizki Duwinanto, Muhammad Indonesia Final Project Convolutional Neural Network, Steganalysis, MP3, Classification INSTITUT TEKNOLOGI BANDUNG https://digilib.itb.ac.id/gdl/view/39479 Steganography is a method for hiding secret messages on media cover in the form of text, audio, images or videos, so that the message is not suspected by those who are not authorized to open the message. The technique to find out whether the cover media is a stego file or not is steganalysis. In this study, the detection of hidden messages focused on MP3 files inserted by the MP3Stego and Huffman Code Mapping algorithms with an embedding rate of 0.1 to classify based on the steganographic algorithm used and the estimated length of the inserted message. In conducting this research, it is necessary to know the audio features of MP3, build suitable deep learning methods and the performance of the models that have been produced. The proposed solution for both problems is to use the QMDCT audio feature and deep learning architecture with Convolutional Neural Network that uses 3 × 3 convolutional layers, 1 × 1 convolutional kernel, activation function, max pooling layer, fully-connected layers, batch normalization layers , adam optimization function and cross entropy log loss. The results of this study are the best algorithm classification model with an accuracy performance of 91.78% with F1-Score 92.22% and the best classification model for message length estimation has an accuracy performance of 24.16% with F1-Score 21.40%. Thus the proposal of deep learning features and architecture is good in classifying algorithms and covers, but still poor in classifying the estimated length of the message. text |
institution |
Institut Teknologi Bandung |
building |
Institut Teknologi Bandung Library |
continent |
Asia |
country |
Indonesia Indonesia |
content_provider |
Institut Teknologi Bandung |
collection |
Digital ITB |
language |
Indonesia |
description |
Steganography is a method for hiding secret messages on media cover in the form of
text, audio, images or videos, so that the message is not suspected by those who are not
authorized to open the message. The technique to find out whether the cover media is a stego
file or not is steganalysis. In this study, the detection of hidden messages focused on MP3 files
inserted by the MP3Stego and Huffman Code Mapping algorithms with an embedding rate of
0.1 to classify based on the steganographic algorithm used and the estimated length of the
inserted message. In conducting this research, it is necessary to know the audio features of
MP3, build suitable deep learning methods and the performance of the models that have been
produced. The proposed solution for both problems is to use the QMDCT audio feature and
deep learning architecture with Convolutional Neural Network that uses 3 × 3 convolutional
layers, 1 × 1 convolutional kernel, activation function, max pooling layer, fully-connected
layers, batch normalization layers , adam optimization function and cross entropy log loss. The
results of this study are the best algorithm classification model with an accuracy performance
of 91.78% with F1-Score 92.22% and the best classification model for message length
estimation has an accuracy performance of 24.16% with F1-Score 21.40%. Thus the proposal
of deep learning features and architecture is good in classifying algorithms and covers, but still
poor in classifying the estimated length of the message. |
format |
Final Project |
author |
Rizki Duwinanto, Muhammad |
spellingShingle |
Rizki Duwinanto, Muhammad AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
author_facet |
Rizki Duwinanto, Muhammad |
author_sort |
Rizki Duwinanto, Muhammad |
title |
AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
title_short |
AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
title_full |
AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
title_fullStr |
AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
title_full_unstemmed |
AUDIO MP3 STEGANALYSIS WITH CONVOLUTIONAL NEURAL NETWORK |
title_sort |
audio mp3 steganalysis with convolutional neural network |
url |
https://digilib.itb.ac.id/gdl/view/39479 |
_version_ |
1822269269322235904 |