Research Article
Hyper-Tuned CNN Using EVO Technique for Efficient Biomedical Image Classification
Table 1
Summary of the state of art models.
| Ref. | Model used | Advantage | Disadvantage |
| [1] | Hyper-tuned CNN using spotted hyena optimizer (SHO) | Improved accuracy 99.25% | Hyperparameter tuning model is considered complex | [2] | CNN and SVM | Accuracy 98.49%. | High computation complexity | [3] | Feedforward back-propagation artificial neural network (FP-ANN) and k-nearest neighbour (k-NN) | FP-ANN and k-NN are 97% and 98% respectively | No hyperparameters optimization | [14] | Hyper-tuned CNN using univariate dynamic encoding (UDE) algorithm. | Above 95% accuracy. | High implementation requirements | [15] | CNN optimized hyperparameters based on genetic algorithm | 90% accuracy | Low accuracy and high computational complexity | [16] | Hyper-tuned parameters of (CNN) using multi-level particle swarm optimization (MPSO). | Above 95% accuracy | The complexity and performance depend on the dimensions of the search space. | [20] | Deep convolutional neural network (DCNN). | 97.72% accuracy | High computation complexity | Wang et al. [18] | CNN with adaptive momentum optimization | 95% accuracy | Low accuracy and high computational complexity |
|
|