Andere Kunden interessierten sich auch für
![LSTM Recurrent Neural Networks for Signature Verification]( "LSTM Recurrent Neural Networks for Signature Verification")
LSTM Recurrent Neural Networks for Signature Verification32,99 €![Forecast Stock Index using Neural Networks and Evolutionary Computing]( "Forecast Stock Index using Neural Networks and Evolutionary Computing")
Forecast Stock Index using Neural Networks and Evolutionary Computing32,99 €![Knowledge Extraction with Neural Networks]( "Knowledge Extraction with Neural Networks")
Knowledge Extraction with Neural Networks44,99 €![Fundamental Simulation Methodologies Dwelling in Neural Networks]( "Fundamental Simulation Methodologies Dwelling in Neural Networks")
Fundamental Simulation Methodologies Dwelling in Neural Networks38,99 €![Towards Optimally Diverse Randomized Ensembles of Neural Networks]( "Towards Optimally Diverse Randomized Ensembles of Neural Networks")
Towards Optimally Diverse Randomized Ensembles of Neural Networks36,99 €![Multistage Neural Networks]( "Multistage Neural Networks")
Multistage Neural Networks32,99 €![Adversarial Multimedia Forensics]( "Adversarial Multimedia Forensics")
Adversarial Multimedia Forensics126,99 €
This book is focused on the revealing of hidden information present in multimedia files, mainly in pictures. This hidden information (messages) is coded in by means of steganography, which is an additional method of cryptography. Steganography provides better security for messages and the detection of such a message is not easy. The main goal of this research is a classification by means of artificial neural networks aimed at reducing false positive classification results to a minimum. To accomplish the main goal, a new model of image pre-processing was proposed. This pre-processing model is based on the Huffman coding, which is the main part of the lossless compression algorithm used in JPEG images. The Huffman coding can be easily transformed into the training sets for the artificial neural network, which is used as a classifier. The type of used artificial neural network was feed forward with supervision and Levenberg-Marquardt training algorithm. The results from performed simulations proved that neural networks are capable of solving such a complex tasks and the border the error in classification was under 1% which is classified as a suitable and powerful tool for steganalysis