KDD on CIFAR 10 Dataset by Deep Learning Models

Not only did conventional ANNs and ML struggle to train models and extract features from large photos, but they also performed poorly when applied to picture classification tasks, leading to inaccurate results. Because of this, the authors of this work set out to provide the groundwork for picture recognition and classification on massive datasets by presenting a deep learning model of picture classification. This research identified the diagnostic techniques that produced the most accurate results for CIFAR10 image classification by deep learning models such as Xception, VGG16, ResNet50, InceptionV3, MobileNet, and DenseNet121 are the most widely implemented. In comparison to other Deep Learning models, the ResNet50 has performed admirably, according to this study. In comparison to other models, this one yields the most favorable results with the fewest deviations. In contrast, the DenseNet121 model exhibits the worst performance (74.50 % accuracy, 0.75 precision, 0.47 recall, 0.75 ROC, 0.69 PRC, 0.44 kappa, 0.25 F-Measure, 0.44 MCC) with the greatest number of deviations. This study investigates the ResNet50 model in light of its results.