Deep Learning (ML II)

About This Course

In this module, students learn how to interpret, modify, and design deep learning architectures. We analyze structural layers, activation functions, and optimization pathways, establishing a classification of architectural principles to solve real-world problems.

A central focus is understanding network training mechanics, hyperparameter optimization, and sequence processing using advanced self-learning principles.

Learning Objectives

By the end of this course, students will be able to:

• Analyze the paradigm shift from classical machine learning to deep neural networks and identify where classical techniques fall short.
• Construct and configure core components of Deep Neural Networks, including input, hidden, and output layers.
• Implement optimization techniques, including backpropagation, evaluation metrics, and hyperparameter tuning.
• Develop Convolutional Neural Networks (CNNs) utilizing filters, kernels, and pooling layers for visual data tasks.
• Build Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM &BI_LSTM) networks for univariate and multivariate time-series forecasting.
• Evaluate Transformer models by understanding their history, structural internals, and multi-head self-attention mechanisms.

Requirements

Students should have prior knowledge of basic machine learning concepts, proficiency in Python, and familiarity with foundational deep learning techniques. Experience with frameworks like TensorFlow or PyTorch is recommended but not required.

General Information

• Teaching Format: Experience
• Total Workload Master: 125h (40h/85h) / 5 ECTS
• Total Workload MBA: 100h (40h/60h) / 4 ECTS
• Total Workload Micro Degree: 125h (40h/85h) / Equivalent to 5 ECTS
• Module coordinator: Prof. Dr. Raad Bin Tareaf
• Examinations: Quizzes, presentation(s), essay(s)/paper(s), project report(s), written exam (tbd) - Details will be announced with course start.
• Offered: Even quarters