Misfire Detection in Spark Ignition Engine Using Transfer Learning

Table 7

Performance of pretrained models for various batch sizes.

Pre-trained model	Classification accuracy for different batch sizes (%)					Overall accuracy (%)
	8	10	16	24	32
AlexNet	89.30	94.00	79.30	89.30	90.70	88.52
VGG-16	98.40	98.70	97.70	97.30	98.30	98.08
GoogLeNet	98.00	98.10	91.00	95.00	94.00	95.20
ResNet-50	96.00	97.60	83.20	93.60	97.40	93.56

Bolded values represent the best classification accuracy obtained by the individual pretrained networks for experimentation with different batch sizes. The pretrained network with best overall classification accuracy is represented in the final column.