Enhanced Convolutional Neural Networks for Improved Image Classification
Xiaoran Yang, Shuhan Yu, Wenxi Xu
TL;DR
The paper tackles CIFAR-10 image classification by designing an enhanced CNN with deeper convolutional blocks, batch normalization, and dropout to improve generalization on a small-scale dataset. The architecture—three conv blocks with two 3x3 convs each, BN after every conv, max-pooling, and 25% dropout—achieves 84.95% test accuracy, outperforming classic baselines like LeNet-5 and standard CNNs. An ablation study demonstrates the distinct contribution of batch normalization, dropout, and network depth to performance, underscoring the value of carefully balanced architectural refinements. The findings suggest that refined CNNs can effectively handle small, low-resolution datasets and motivate extensions to CIFAR-100 and transfer learning to larger datasets for broader applicability.
Abstract
Image classification is a fundamental task in computer vision with diverse applications, ranging from autonomous systems to medical imaging. The CIFAR-10 dataset is a widely used benchmark to evaluate the performance of classification models on small-scale, multi-class datasets. Convolutional Neural Networks (CNNs) have demonstrated state-of-the-art results; however, they often suffer from overfitting and suboptimal feature representation when applied to challenging datasets like CIFAR-10. In this paper, we propose an enhanced CNN architecture that integrates deeper convolutional blocks, batch normalization, and dropout regularization to achieve superior performance. The proposed model achieves a test accuracy of 84.95%, outperforming baseline CNN architectures. Through detailed ablation studies, we demonstrate the effectiveness of the enhancements and analyze the hierarchical feature representations. This work highlights the potential of refined CNN architectures for tackling small-scale image classification problems effectively.