Akademik Dersler - İçerik

Ability to explain the layer architectures of artificial neural networks and convolutional neural networks (CNN), the theoretical structure of convolution operations, and the working principles of state-of-the-art deep learning models.
Ability to apply image preprocessing steps for deep learning models; and to analyze problems encountered in imbalanced datasets and deliver solutions using SMOTE and various data augmentation techniques.
Ability to design, train, and implement end-to-end engineering applications of custom deep learning architectures using publicly available datasets.
Ability to quantitatively evaluate trained models using appropriate performance metrics; and to visualize and explain the model's decision-making mechanism using Class Activation Mapping (CAM/Grad-CAM) methods.