Mohammad Fakhredanesh

Haniyeh Rafiee, Mohammad Fakhredanesh

In this article, one of the most important methods of steganography on VoIP called echo hiding is improved. This method has advantages in maintaining the statistical and perceptual characteristics of audio signals as well as security against the sensitivity of the human audio system (HAS). However, it has lots of errors in detecting coded and hidden messages, which is detectable using existing steganalysis methods. The percentage of extracting messages in these improved methods of echo hiding is high, but they lower the security of the method. In this article, a method is presented to improve the method of extracting echo hiding, and enhance its security through a combined method based on spread spectrum. To improve the extraction, a wrong hypothesis is corrected and substituted. To improve security using a pseudo-random key generation algorithm, spread spectrum and echo hiding methods are used randomly. To evaluate the proposed extraction, numerous extraction tests are carried out in the normal state and in the event of attacks. A steganalyser has also been used to assess security improvements. The results gained through different experiments on the security of steganography indicate a 3-percent increase in steganalysis errors. The proposed extraction method was modified based on the main method and resulted in more than 10% improvement.

Mohammad Ebrahim Khademi, Mohammad Fakhredanesh, Seyed Mojtaba Hoseini

Traditional methods of summarization are not cost-effective and possible today. Extractive summarization is a process that helps to extract the most important sentences from a text automatically and generates a short informative summary. In this work, we propose an unsupervised method to summarize Persian texts. This method is a novel hybrid approach that clusters the concepts of the text using deep learning and traditional statistical methods. First we produce a word embedding based on Hamshahri2 corpus and a dictionary of word frequencies. Then the proposed algorithm extracts the keywords of the document, clusters its concepts, and finally ranks the sentences to produce the summary. We evaluated the proposed method on Pasokh single-document corpus using the ROUGE evaluation measure. Without using any hand-crafted features, our proposed method achieves state-of-the-art results. We compared our unsupervised method with the best supervised Persian methods and we achieved an overall improvement of ROUGE-2 recall score of 7.5%.