Search

Nowadays, machine learning and deep learning have proven to be an encouraging field after artificial intelligence. Deep learning being operative, supervised, time and cost-effective technique evolving has its myriad applications not just in the field of pattern recognition and prediction but also can be utilized for addressing some important issues in the field of data science and data analytics. Deep learning came into existence with the availability of the exabyte of data collected with time. This subset technique of machine learning acquires the instructive, differential, and distinct features of data. Deep learning is a subset of machine learning that has made a noteworthy breakthrough as it is inspired by the working of the brain and its cells called neurons now coming up with significant performance, efficient evaluation and remarkable outcomes in a huge range of applications, and deep learning techniques have set their way with useful security tools and are now widely used in the field of statistics, neuroscience, data analytics, pattern recognition, image sensing, and the list goes on. This paper covers the basics of deep learning, its several commonly used algorithms and techniques, their architectures along with several application areas where these techniques have shown prominent differences and revolutionary change with the time. We have also tried to generalize the characteristics of different used methods of deep learning. At last, we present the challenges for the growth of these methods, concluding with our perspectives and future scope.

Automated Face Recognition involves the extraction of features of the facial images of an unknown person and the identification by comparing the features with the available feature set of a known person. Each facial image has been divided into 16 independent sub-blocks which have been used in this work for designing and implementation of the Markov Random Field- (MRF) based classifier which helps to generate the Maximum A Posteriori (MAP) probability. Using the relationship of MRF and Gibbs Distribution, the MAP has been converted into an energy function. A novel approach for defining the energy function has also been proposed in this work.

One of the major concepts of voting is alternative. There are numerous examples of alternatives, e.g., several versions of an immigration reform bill, electing the political party, etc. In this paper, we are considering the survey result, which shows the preferences over different mobile service provider’s alternatives. Statistical data referred from TelecomLead (TelecomLead—Magazine. https://www.telecomlead.com, [1]) choosing mobile service provider’s customer share playing a major role in Indian Telecommunication scenario. We use here Plurality, Copeland, and Borda’s rule and tried to find the correlation between MSP’s.

Modern malware detection systems have largely relied on the definition of signatures to characterize malwares to their corresponding malware families. These signatures that characterize malware families are parts of codes and it is believed that families of malwares share commonalities in their signatures. We hypothesize that changes in these signatures generate newer sub-families of malwares. In the present work we have evaluated the signature conservation across two sub-families of rootkits. We have carried out our experiments to establish the fact that features in the rootkit family of malware are conserved. We report that our feature extraction yielded the accuracy of 84.17% using the Naive Bayes classification algorithm. The results reported in this work reinforce our belief that there are subsets of independent features that discriminate between sub-families but not exhibiting any trend of conservation. We conclude that certain features (if not all) are preserved and discriminate between sub-families.

Data is nothing but information of anything and we know it will continue to grow more and more. Unspecified format of data is unstructured data known as big data. 25% of the data that exist is in specified format, i.e., structured data and other 75% is in unspecified format. Unstructured data can be found anywhere. Generally, most of people and organizations pass out their lives working around unstructured data. In this paper, we have tried to work on how one can store unstructured data.