Search

Health care institutions generated large volume health care data which are collected from the numerous patients. These health care data consists of various hidden and useful information about the different types of diseases which needs to be extracted accurately for the patient treatment. However analysis big health care data becomes more difficult task due to its excess volume and variety of data. There is various research methodologies has been proposed earlier to analyse and extract the useful information from the health care big data. In this analysis work, discussion about the different research methods has been given whose main goal is to extract the useful information from the health care big data. This analysis work provides the overview of various big data analytics methods which are applied on health care information to extract the useful information. This work provides the detailed description about the working procedure of different research methodologies and also denoted the merits and demerits of various research methods. The comparison evaluation of the different research techniques has been performed based on merits and demerits from which it is concluded the best research method as for now that can ensure the reliable and accurate analysis on health care big data. [ABSTRACT FROM AUTHOR]

Botnets are becoming an easy way of creating multiple attacks. One of them was botnets simulate the behaviour that is very near to the legitimate users. Previous research found through semi-Markov model it was difficult to detect mimicking attack based on browsing statistics if attacking bots were sufficiently large in number [1]. By using Bots attackers will collect the user profiles from multiple systems. Bot master (attacker) will study statistics and Bot master will prepare a common profile (or) multiple profiles similar to the user activities. In the next phase, bot master injects profile into user systems through bots. If bot master injects common profile across all bot injected system then the attack was considered as a homogeneous mimicking attack. If bot master injects multiple profiles to the bot injected systems the attack was considered a heterogeneous mimicking attack. As part of our study, we simulated the mimicking attack and worked on detecting at multiple levels. We have developed algorithms of detecting at a server level [2] and the gateway level [3]. In this paper, we are going to discuss the merits and demerits of these two detection algorithms and proposing another architecture module hybrid level detection. Which will be spread across servers and gateway which will have the bird view of activities happening in the network. It collects the statistics from different network elements and based on the analysis of the trace of the bot activities will identify mimicking attack. [ABSTRACT FROM AUTHOR]

In the websites the contents will be are similarity when we compared with other search engines. So to check the similar content in the websites and its web contents we created a overhead to the search engine which will severely effect its performance & quality. So to detect the silmilar or same content or web documenattion some techniques are implemented by web crawling research community. So it is one of major factor for the search engines to provide some applicatory data to users in the first page itself. So to avoid such issues we proposed a methodlogy called Automatic Detection of illegitimate websites with Mutual Clustering (ADIWMC) paper we are presenting a peculiar and efficacious path for the detection of similarities in the web pages in web clustering. Detection of same and similar web pages and web content will be done by storing the crawled web pages into depository. Initially the adwords will be extracted from the crawled pages and similarity checking will be done between the two pages based in the usage of adwords. So a threshold value is set for this, if the similarity checking percentage is greater than the threshold then similarity content is reduced and improves the depositary and improves the search engine quality. In the sections of existing analysis and the proposed analysis we are clearly exploring how it works. [ABSTRACT FROM AUTHOR]

In this paper we propose a novel edge detection technique. Edges and noise pixels exhibits similar characteristics, unfortunately most of the existing techniques fails in generating the satisfactory results because they are not satisfying the criterion parameters of qualitative evaluation. There by, we introduce Gray Level Spatial Correlation technique based on a statistical parameter computed as the absolute difference between Global mean of entire image and local mean of a 3X3 map. The resultant similarity values other than 9 evident object edges; hence this technique eliminates most of the noise pixels and highlights the object edges. This approach can be further extended to describe the misclassification region or Region of Interest (ROI) for many supervised segmentation and threshold estimation methods. The results obtained are very promising in comparison with standard existing techniques. This method converges at low time complexities. [ABSTRACT FROM AUTHOR]