Intrusion Detection Techniques in Social Media Cloud: Review and Future Directions

Table 2

Classification of different intrusion detection techniques.

Technique	Definition	Examples	References
Statistical methods	Statistical methods deal with fewer amounts of data and require more human effort	(i) Estimating critical network parameter thresholds; (ii) PCA, chi-square distribution, and Gaussian mixture distribution; (iii) Least square support vector machine (LS-SVM); (iv) Genetic algorithm (GA); (v) Hybrid classifier (DTNB), which is a combination of decision table (DT) and NB algorithms	(i) [26]; (ii) [27]; (iii) [28]; (iv) [29]; (v) [30]
AI and machine learning methods	AI and ML methods have a high predictive capacity and need little human intervention	(i) Naïve Bayesian classifier and a decision tree; (ii) Naïve Bayesian classifier; (iii) SVM; (iv) Hybrid model that incorporates data mining approaches such as the -means clustering algorithm and the RBF kernel function of the support vector machine; (v) Decision tree (J48) algorithm; (vi) RF; (vii) C4.5 decision tree; (viii) -means and -nearest neighbors; (ix) GA and best feature set selection (BFSS); (x) RF and SVM	(i) [31]; (ii) [32]; (iii) [33]; (iv) [34]; (v) [35]; (vi) [36]; (vii) [37]; (viii) [38]; (ix) [39]; (x) [40]
Swarm intelligence and evolutionary computation methods	It uses GAs, particle swarm optimization (PSO), and differential evolution computational techniques	(i) GAs, PSO, and differential evolution (DE); (ii) Particle swarm optimization classifiers with genetic-particle swarm optimization; (iii) Feature selection (EFS) algorithm and teaching learning-based optimization (TLBO) methodologies are combined; (iv) Combines the approaches of PSO, GA, and neural networks (RBF)	(i) [41]; (ii) [42]; (iii) [43]; (iv) [44]
Deep learning methods	Part of AI where the model is a mathematical algorithm that is trained to provide the same conclusion or prediction as a human expert	(i) Self-taught learning (STL); (ii) Deep convolutional neural network (DCNN); (iii) The deep belief network (DBN) (stacking multiple RBMs); (iv) DCNN; (v) DCNN; (vi) Long short-term memory (LSTM) and DBN; (vii) Deep neural network, which consists of sparse autoencoder and logistic regression. By stacking the autoencoders, a deep network is built, and a logistic regression network is used to classify the features learned; (viii) DCNN; (ix) DCNN; (x) DCNN; (xi) SVM and DCNN; (xii) DCNN; (xiii) Deep convolutional neural networks (CNNs); (xiv) LSTM; (xv) CNN; (xvi) CNN	(i) [45]; (ii) [46]; (iii) [47]; (iv) [48]; (v) [49]; (vi) [50]; (vii) [51]; (viii) [52]; (ix) [53]; (x) [54]; (xi) [55]; (xii) [56]; (xiii) [57]; (xiv) [58]; (xv) [59]; (xvi) [60]