Introduction
-
Introduction of the notion of one-shot learning for efficient detection of COVID-19 chest X-ray images.
-
Proposal of ensembling of GRNN and PNN classifiers using AND operation, which works on one-shot cluster-based learning to cope with the present pandemic situation.
-
Exhaustive experimentation and comparative analysis of the results obtained against that of deep learning models.
Proposed Pipeline
One-shot Learning and Feature Representation
Neural Network-Based Classifiers
Experimentation and Comparative Analysis
Dataset
Experimental Setup
Cluster and Ensemble Approach
Methods | Class | COVID-19 | Normal | Pneu. Bac. | Pneu. Vir | Average | Train | Test |
---|---|---|---|---|---|---|---|---|
Alexnet [31, 32] | 2 | 100 | 100 | – | – | 100 | 130 | 18 |
3 | 100 | 77.7 | 77.8 | – | 85.1 | 200 | 27 | |
4 | 100 | 64.3 | 44.4 | 50 | 64.67 | 270 | 36 | |
Googlenet [31, 34] | 2 | 100 | 100 | – | – | 100 | 130 | 18 |
3 | 81.8 | 75 | 87.5 | – | 81.4 | 200 | 27 | |
4 | 100 | 100 | 70 | 66.7 | 84.0 | 270 | 36 | |
Resnet18 [31, 33] | 2 | 100 | 100 | – | – | 100 | 130 | 18 |
3 | 100 | 100 | 64.3 | – | 81.4 | 200 | 27 | |
4 | 100 | 100 | 50 | 40 | 72.5 | 270 | 36 | |
Proposed method | 2 | 100 | 100 | – | – | 100 | 04 | 148 |
Cluster 1 (5 samples) | ||||||||
3 | 100 | 64.55 | 68.35 | – | 77.61 | 5 | 227 | |
Cluster 2 (7 samples) | ||||||||
3 | 100 | 70.99 | 68.35 | – | 79.76 | 7 | 227 | |
Cluster 3 (13 samples) | ||||||||
3 | 100 | 92.49 | 63.29 | – | 85.23 | 13 | 227 | |
Cluster 1 (4 samples) | ||||||||
4 | 94.2 | 54.43 | 54.43 | 44.3 | 61.84 | 4 | 306 | |
Cluster 2 (5 samples) | ||||||||
4 | 100 | 54.43 | 54.43 | 44.3 | 63.29 | 5 | 306 | |
Cluster 3 (6 samples) | ||||||||
4 | 100 | 58.23 | 53.16 | 43.04 | 63.60 | 6 | 306 | |
Cluster 4 (6 samples) | ||||||||
4 | 100 | 69.62 | 51.90 | 41.77 | 65.80 | 6 | 306 | |
Cluster 5 (29 samples) | ||||||||
4 | 100 | 86.08 | 65.82 | 44.30 | 74.05 | 29 | 306 |