| ARTIFICIAL INTELLIGENCE ENGINEERING | |||||
| Bachelor | TR-NQF-HE: Level 6 | QF-EHEA: First Cycle | EQF-LLL: Level 6 | ||
| Course Code | Course Name | Semester | Theoretical | Practical | Credit | ECTS |
| AIN3002 | Deep Learning | Spring | 3 | 0 | 3 | 6 |
| Language of instruction: | English |
| Type of course: | Must Course |
| Course Level: | Bachelor’s Degree (First Cycle) |
| Mode of Delivery: | Hybrid |
| Course Coordinator : | Assist. Prof. FATİH KAHRAMAN |
| Recommended Optional Program Components: | - |
| Course Objectives: | Deep Learning (DL) is a subfield of Machine Learning (ML) that focuses on the development and training of artificial neural networks to automatically learn hierarchical patterns and representations from data. Over the past decade, Deep Learning has revolutionized the field of artificial intelligence (AI), enabling groundbreaking progress in areas such as computer vision, natural language processing, speech recognition, and generative modeling. By leveraging large-scale datasets, advanced architectures, and high-performance computing, DL models have demonstrated remarkable capabilities in solving complex, real-world problems with human-level or even superhuman performance. This course provides an in-depth exploration of Deep Learning, combining advanced theoretical concepts with practical implementation using industry-standard tools like TensorFlow and PyTorch, which are also widely used in academic research. Through a mix of lectures and hands-on coding sessions, students will cover a broad range of topics including neural network architectures, optimization strategies, regularization techniques, convolutional and recurrent neural networks, transformers, generative models, and selected state-of-the-art architectures. To support further hands-on learning, we will provide access to carefully selected open-source online resources and tutorials, along with assignments designed to reinforce coding skills and project development. As part of the course requirements, students will complete a comprehensive final project in which they apply deep learning techniques to solve a real-world problem of their choice. The project will culminate in a class presentation, during which students will receive constructive feedback from both the instructor and their peers to enhance their understanding and communication of deep learning applications. This course will primarily use the Python programming language, with TensorFlow and PyTorch as the main frameworks for coding. |
|
The students who have succeeded in this course; I. To understand the foundation models II. To apply the algorithms to real problems III. To design and train neural networks IV. To explore advanced deep learning architectures V. To understand a state-of-the-art DL learning algorithms |
| This Deep Learning course, as a fundamental component of AI, plays a crucial role in developing the essential skills and competencies required in the Artificial Intelligence program. By bridging the gap between theoretical knowledge and practical application, the course significantly contributes to the vocational education and career readiness of students, preparing them for roles in AI, machine learning, deep learning, and data science. The course learning outcomes directly align with the program's overall objectives by fostering a deeper understanding of AI technologies and their real-world applications. Students will develop advanced technical skills, including the design, implementation, and optimization of deep learning models, which are essential for solving complex problems in diverse fields such as computer vision and natural language processing Additionally, the course emphasizes practical coding experience with industry-standard tools like TensorFlow and PyTorch, ensuring students are equipped with hands-on skills that are highly valued in the job market. Furthermore, the course supports the program's goal of producing graduates who are capable of conducting independent research and tackling advanced problems in AI. Students will enhance their problem-solving abilities through project-based learning, where they apply deep learning techniques to real-world challenges. By fine-tuning deep learning models, evaluating their performance, and understanding hyperparameters, students will gain the expertise necessary to optimize and deploy AI solutions efficiently. |
| Week | Subject | Related Preparation |
| 1) | Overview of AI, ML, and DL; Historical Evolution of DL; Efficient DL; DL and AI Applications in Different Domains; Course Principles | |
| 2) | Mathematical foundations: Derivatives, Matrices; Basics of neural networks: perceptron, logistic regression | |
| 3) | Mathematical foundations: Derivatives, Matrices; Basics of neural networks: perceptron, logistic regression Hyperparameters; Activation Functions; Loss Functions; Optimization and Learning Rate | |
| 4) | Common Optimization Techniques; Dataset Split for Training and Test; Overfitting and Underfitting | |
| 5) | Regularization Techniques; Introduction on Google Colab and Coding with Tensorflow and Pytorch; Applying Linear Regression by Tensorflow and Pytorch in Google Colab | Practice on the provided code notebooks for hyperparameter tuning. Course 1 from Coursera Delivery: submission before 5th session |
| 6) | Image Fundamentals; Computer Vision and Applications; Image-based Classification; Convolutional Neural Networks (CNNs); Applying Image Classification by Tensorflow and Pytorch in Google Colab | Practice on the provided code notebooks for hyperparameter tuning |
| 7) | Hyperparameters of Convolutional Layer; Pooling Layer; Batch Normalization; Classification Layers in CNN; Evaluation Metrics for Image Classification | Course 2 from Coursera Delivery: submission before 7th session |
| 8) | Midterm Exam | |
| 9) | ROC Curve as an Evaluation Metric; Common CNN Models: LeNet, AlexNet, VGG, GoogLeNet, ResNet; Transfer Learning; Applications of CNN; Applying CNN models by Tensorflow and Pytorch in GoogleColab | - Course 3 from Coursera - Project topic and proposal Delivery: submission before 9th session |
| 10) | Recurrent Neural Network; Applying RNN models by Tensorflow and Pytorch in Google Colab | Practice on the provided code notebooks for hyperparameter tuning |
| 11) | Transformers; Applying Transformer models by Tensorflow and Pytorch in Google Colab | Course 4 from Coursera Delivery: submission before 11th session |
| 12) | Deep Generative Models; Applying generative models by Tensorflow and Pytorch in Google Colab | Practice on the provided code notebooks for hyperparameter tuning |
| 13) | Other Applications and Deep Learning Architectures, New Frontiers, and State-of-the-art models | Course 5 from Coursera Delivery: submission before 13th session |
| 14) | Final review of the course | Student project presentations All components of final project Delivery: submission before 14th session |
| 14) | Final review of the course | Student project presentations All components of final project Delivery: submission before 14th session |
| Course Notes / Textbooks: | Title: “Deep Learning” Author: Ian Goodfellow, Yoshua Bengio, Aaron Courville Link: https://www.deeplearningbook.org/ ➢ Title: “Deep Learning for Computer Vision” Author: Adrian Rosebrock Link: https://pyimagesearch.com/deep-learning-computer-vision-python-book/ ➢ Title: “Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow” Author: Aurélien Géron Link:https://www.amazon.com/Hands-Machine-Learning-Scikit-LearnTensorFlow/dp/1492032646 ➢ Title: “Neural Networks and Deep Learning” Author: Michael Nielsen Link: http://neuralnetworksanddeeplearning.com/ ➢ Title: “Deep Learning with Python” Author: François Chollet Link: https://www.amazon.com/Deep-Learning-Python-Fran%C3%A7ois-Chollet/dp/1617296864 |
| References: | Title: “Deep Learning” Author: Ian Goodfellow, Yoshua Bengio, Aaron Courville Link: https://www.deeplearningbook.org/ ➢ Title: “Deep Learning for Computer Vision” Author: Adrian Rosebrock Link: https://pyimagesearch.com/deep-learning-computer-vision-python-book/ ➢ Title: “Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow” Author: Aurélien Géron Link:https://www.amazon.com/Hands-Machine-Learning-Scikit-LearnTensorFlow/dp/1492032646 ➢ Title: “Neural Networks and Deep Learning” Author: Michael Nielsen Link: http://neuralnetworksanddeeplearning.com/ ➢ Title: “Deep Learning with Python” Author: François Chollet Link: https://www.amazon.com/Deep-Learning-Python-Fran%C3%A7ois-Chollet/dp/1617296864 |
| Semester Requirements | Number of Activities | Level of Contribution |
| Quizzes | 5 | % 10 |
| Homework Assignments | 4 | % 10 |
| Project | 1 | % 20 |
| Midterms | 1 | % 20 |
| Final | 1 | % 40 |
| Total | % 100 | |
| PERCENTAGE OF SEMESTER WORK | % 40 | |
| PERCENTAGE OF FINAL WORK | % 60 | |
| Total | % 100 | |
| Activities | Number of Activities | Duration (Hours) | Workload |
| Course Hours | 14 | 3 | 42 |
| Study Hours Out of Class | 14 | 3 | 42 |
| Presentations / Seminar | 1 | 2 | 2 |
| Project | 1 | 12 | 12 |
| Homework Assignments | 4 | 2 | 8 |
| Quizzes | 4 | 1 | 4 |
| Midterms | 1 | 2 | 2 |
| Final | 1 | 2 | 2 |
| Total Workload | 114 | ||
| No Effect | 1 Lowest | 2 Low | 3 Average | 4 High | 5 Highest |
| Program Outcomes | Level of Contribution | |
| 1) | Build up a body of knowledge in mathematics, science and Artificial Intelligence Engineering subjects; use theoretical and applied information in these areas to model and solve complex engineering problems. | 5 |
| 2) | Design complex Artificial Intelligence systems, platforms, processes, devices or products under realistic constraints and conditions, in such a way as to meet the desired result; apply modern design methods for this purpose. | 5 |
| 3) | Identify, formulate, and solve complex Artificial Intelligence Engineering problems; select and apply proper modeling and analysis methods for this purpose. | 4 |
| 4) | Devise, select, and use modern techniques and tools needed for solving complex problems in Artificial Intelligence Engineering practice; employ information technologies effectively. | 5 |
| 5) | Design and conduct numerical or physical experiments, collect data, analyze and interpret results for investigating the complex problems specific to Artificial Intelligence Engineering. | 5 |
| 6) | Ability to communicate effectively in English and Turkish (if he/she is a Turkish citizen), both orally and in writing. Write and understand reports, prepare design and production reports, deliver effective presentations, give and receive clear and understandable instructions. | 5 |
| 7) | Recognize the need for life-long learning; show ability to access information, to follow developments in science and technology, and to continuously educate oneself. | |
| 8) | Develop an awareness of professional and ethical responsibility, and behave accordingly. Be informed about the standards used in Artificial Intelligence Engineering applications. | |
| 9) | Learn about business life practices such as project management, risk management, and change management; develop an awareness of entrepreneurship, innovation, and sustainable development. | |
| 10) | Acquire knowledge about the effects of practices of Artificial Intelligence Engineering on health, environment, security in universal and social scope, and the contemporary problems of Artificial Intelligence Engineering; is aware of the legal consequences of Mechatronics engineering solutions. | |
| 11) | Cooperate efficiently in intra-disciplinary and multi-disciplinary teams; and show self-reliance when working on Artificial Intelligence-related problems. |