|
Reference | Purpose | Result |
|
[37] | BDA helped in searching for the optimal parameter sets (kernel parameter and penalty factor) for KELM and the optimal feature subset among the feature candidates simultaneously | BDA showed its superiority as a searching technique to find the set of optimal parameters and the optimal feature subset |
|
[40] | Multilevel segmentation of colour fundus images | Using the DA as an optimization algorithm produced better results for segmenting colour images |
|
[41] | In a watermarking technique for the medical images, DA was utilized to select the effective pixels | The correlation coefficient values using the DA were greater than the other techniques such as PSO, GA, and random selection |
|
[42] | Exploring the pixels of images and discovering which pixel contains significant information about the object (DA was used as a detection model) | The DA could work as an efficient and fast object extraction from images |
|
[43] | DA was used as a parameter optimizer of SVM; furthermore, the effect of the number of solutions and generations on the accuracy of the produced result and computation time was investigated | It was shown that the classification error rate for the proposed work was lower than that in PSO + SVM and GA + SVM, and the reason for this was that the DA parameters could be altered iteratively; furthermore, it was shown that increasing either the number of solutions or generations decreased the rate of misclassification and rose the computational time |
|
[47] | New updating mechanism and elitism were added to the binary dragonfly algorithm; the improved technique was then used to classify different signal types of infant cry, and it was used to overcome the dimensionality problem and select the most salient features | It was noted that the improved technique reduced the percentage of error rate compared with the original binary dragonfly algorithm |
|
[48] | The DA-based artificial neural network technique was utilized for predicting the primary fuel demand in India | The proposed model using the DA was provided with more accurate results comparing to the existing regression models |
|
[49] | Binary-BDA, multi-BDA, and ensemble learning-based BDA were used for wavelength selection | Using binary-BDA causes instability; however, stability boosted by using the multi-BDA and the ensemble learning-based BDA; in addition, the computational complexity of ensemble learning-based BDA was lower than the multi-BDA |
|
[50] | Instead of gradient-based techniques, DA was used for designing filters of IIR | Using the DA prevented trapping into local optima and coefficients close to the actual value were evaluated, and the minimum mean square value was found; in addition, the superiority of the DA was proved to compare to the PSO, CSO, and BA for the aforementioned problem |
|
[51] | Dragonfly-based clustering algorithm was used to focus on the scalability of the internet of vehicles | The proposed technique was compared to a comprehensive learning PSO and ant colony optimization algorithm; the results proved that in high density and medium density the examined technique showed better and average performance, respectively; however, in a low density, the proposed technique performance was bad while the comprehensive learning PSO performed well |
|
[52] | Dragonfly algorithm was utilized to predict the location of randomly deployed nodes in a designated area; also it was used to localizing different noise percentages of distance measurement (Pn) | For range-based localization with varying Pn, dragonflies could produce fewer errors compared with PSO; furthermore, increasing Pn caused an increase in the distances between real and approximated nodes by DA and PSO |
|
[53] | DA was used to enlarge the lifetime of the RFID network | The cluster breakage was reduced through choosing the cluster heads that had similar mobility but high leftover energy; this reduction reduced energy consuming; hence, compared with the existing techniques the efficiency was improved |
|
[54] | DA with two selection probabilities were used as new loud balancing technique called (FDLA); the new technique was then used to keep the stability of processing multiple tasks in the cloud environment | The proposed technique provided the minimum load by allocating less number of tasks |
|
[57] | DA was utilized to examine the optimal sizing and location of distributed generation in radial distribution systems to reduce the power loss in the network | Compared with the DA and WOA, MFO performed better and converged earlier |
|
[59] | In the court case assignment problem, the ability of the judicial system highly depends on time and the efficiency of operation in the court case; the DA was used to find the optimal solution of the assignment problem | The DA could show superior results compared with the FA |
|
[60] | DA was used to optimize the optimum sitting of the capacitor in different radial distribution systems (RDSs); the main aim of this study was to minimize power loss and total cost with voltage profile enhancement | The results proved that DA-based optimization provided comparative results with GWO- and MFO-based optimization methods in terms of a small number of iterations and convergence time; however, it provided superior results compared with the PSO-based technique |
|