Image Classification through Deep Learning Models using Machine Learning Techniques

Large-scale image dataset processing for classification has remained a significant computational burden on the general-purpose Artificial Neural Networks (ANNs) and Classical Machine Learning (ML), resulting in lower computational efficiency and not very high levels of accuracy. In order to overcome these limitations, in this paper, we propose a deep learning framework well optimized for learning image-classification-specific tasks. The paper introduces the basics of neural networks and the importance of different CNN architectures in image classification in the beginning. It makes the following essential improvements to traditional CNN models for improved feature extraction, noise reduction, and tuning of parameters. A custom deep learning network architecture is designed and tuned towards high performance and scalability. Comprehensive experiments also show that the proposed model outperforms a number of state-of-the-art CNN models in classification accuracy with various iterations. Furthermore, the effect of architecture and parameter tuning on performance is deeply investigated. Results demonstrate the efficiency and effectiveness of our proposed model in solving complex image classification problems.