About This Course

The course covers perceptrons, multilayer neural networks, activation and loss functions, gradient descent, and backpropagation. Students study techniques for improving training and generalization, including regularization, dropout, early stopping, and adaptive optimization. Key architectures include convolutional neural networks (CNNs), recurrent networks (RNNs, LSTM, and GRU), attention mechanisms, Transformers, autoencoders, and variational autoencoders. Applications include image analysis, natural language processing, time-series forecasting, anomaly detection, and data generation.

Upon completion of the course, students will be able to explain neural network learning, select suitable architectures and training methods, and evaluate model performance. They will also be able to identify overfitting and training difficulties, apply techniques to improve generalization, and critically assess the capabilities and limitations of deep learning models.