Towards data fusion-based big data analytics for intrusion detection

Intrusion detection is seen as the most promising way for computer security. It used to protect networks against different types of attacks. The major problem in literature classification data into two main classes: normal and intrusion. To solve this problem, several approaches have been proposed but false alarms still present. provide a solution we new intrusion approach based on fusion. objective work suggest an fusion-based Big Data analytics detect intrusions; build one dataset which combines various datasets contains all attack types. This research consists merging heterogeneous removing redundancy information using tools: Hadoop/MapReduce Neo4j. In next step, machine learning algorithms are implemented learning. first algorithm, called SSDM (Semantically Similar Miner), uses fuzzy logic generate association rules between item sets. second K2, score-based greedy search algorithm Bayesian from data. Experimentation results prove that – both cases fusion contributes having very good results.