ISTINYE UNIVERSITY BİLGİ PAKETİ / DERS KATALOĞU
| Data Science (Master) (with Thesis) | |||||
| Master | TR-NQF-HE: Level 7 | QF-EHEA: Second Cycle | EQF-LLL: Level 7 | ||
Course Introduction and Application Information
| Course Code: | VB5004 | ||||
| Course Name: | Deep Learning | ||||
| Semester: | Fall | ||||
| Course Credits: |
|
||||
| Language of instruction: | Turkish | ||||
| Course Condition: | |||||
| Does the Course Require Work Experience?: | No | ||||
| Type of course: | Departmental Elective | ||||
| Course Level: |
|
||||
| Mode of Delivery: | E-Learning | ||||
| Course Coordinator: | Doç. Dr. ŞEBNEM ÖZDEMİR | ||||
| Course Lecturer(s): |
Dr. Öğr. Üy. YASİN KIRELLİ |
||||
| Course Assistants: |
Course Objective and Content
| Course Objectives: | The main objective of this deep learning course is to provide students with advanced knowledge and skills in deep learning theory and algorithms as well as their practical applications. We aim to develop a comprehensive understanding of the mathematical underpinnings of deep learning, different deep learning models, network architectures and how these models can be applied to real-world problems. Furthermore, upon graduation, students are expected to have the capacity to analyze academic or industrial problems and generate stable deep learning solutions. |
| Course Content: | 1. Mathematical foundations and basic concepts of deep learning. 2. Multilayer perceptrons, hyperparameter optimization, and feed-forward networks. 3. Convolutional neural networks and their applications to image processing. 4. Backpropagation algorithm, optimization methods and weighting updates. 5. Long Short Term Memory (LSTM) and recurrent neural networks (RNNs); studies on sequence data. |
Learning Outcomes
|
The students who have succeeded in this course;
1) - Gain the ability to explain deep learning theories and basic concepts. 2) - Ability to build and train deep network architectures. 3) - Develop the ability to apply advanced deep learning algorithms on real-world data sets. 4) - Gain the ability to analyze and improve the performance of deep learning models. 5) - To have the knowledge to develop and apply deep learning models for different problems. |
Course Flow Plan
| Week | Subject | Related Preparation |
| 1) | Introduction to the basic concepts of deep learning and artificial neural networks. | |
| 2) | Mathematical and statistical foundations of deep learning. | |
| 3) | Configuration and training of multilayer perceptrons. | |
| 4) | Feed-forward networks and activation functions. | |
| 4) | Feed-forward networks and activation functions. | |
| 5) | Backpropagation and different optimization algorithms. | |
| 6) | Hyperparameterization and overfitting avoidance methods | |
| 7) | Convolutional Neural Networks (CNN) architectures and applications. | |
| 8) | Midterm Exam | |
| 9) | Advanced concepts and review of CNN network architectures. | |
| 10) | Introduction to RNN and LSTM network structures. | |
| 11) | Array data processing and language processing applications. | |
| 12) | Attention mechanisms and applications in deep learning. | |
| 13) | Autoencoders and generative adversarial networks (GANs). | |
| 14) | Deep Reinforcement Learning. | |
| 15) | Ethical and legal aspects of deep learning models. | |
| 16) | Final Exam |
Sources
| Course Notes / Textbooks: | 1. Goodfellow, I., Bengio, Y., ve Courville, A. (2016). Deep Learning. MIT Press. 2. Chollet, F. (2018). Deep Learning with Python. Manning Publications. |
| References: | 1. LeCun, Y., Bengio, Y., ve Hinton, G. (2015). Deep learning. Nature, 521(7553), 436. 2. Krizhevsky, A., Sutskever, I., ve Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. NIPS. |
Course - Program Learning Outcome Relationship
| Course Learning Outcomes | 1 |
2 |
3 |
4 |
5 |
|---|---|---|---|---|---|
| Program Outcomes | |||||
| 1) Students who successfully complete this program, Knows the scope of technical applications of data science and the tools that can be used. | 3 | 3 | 3 | 3 | 2 |
| 2) Students who successfully complete this program, Knows the effects of application results on society-culture-law. | 2 | 2 | 3 | 2 | 3 |
| 3) Students who complete this program; Recognizes mathematics and code in application processes. | 2 | 3 | 2 | 3 | 3 |
| 4) Students who complete this program; Explain the effects of processes in data science on output and individual. | 3 | 2 | 2 | 2 | 2 |
| 5) Students who successfully complete this program, Understands the insight-foresight and foresight created by data science as a whole in the face of a certain discipline/case. | 2 | 2 | 3 | 3 | 3 |
Course - Learning Outcome Relationship
| No Effect | 1 Lowest | 2 Average | 3 Highest |
| Program Outcomes | Level of Contribution | |
| 1) | Students who successfully complete this program, Knows the scope of technical applications of data science and the tools that can be used. | 2 |
| 2) | Students who successfully complete this program, Knows the effects of application results on society-culture-law. | 3 |
| 3) | Students who complete this program; Recognizes mathematics and code in application processes. | 3 |
| 4) | Students who complete this program; Explain the effects of processes in data science on output and individual. | 2 |
| 5) | Students who successfully complete this program, Understands the insight-foresight and foresight created by data science as a whole in the face of a certain discipline/case. | 2 |
Assessment & Grading
| Semester Requirements | Number of Activities | Level of Contribution |
| Midterms | 1 | % 40 |
| Final | 1 | % 60 |
| total | % 100 | |
| PERCENTAGE OF SEMESTER WORK | % 40 | |
| PERCENTAGE OF FINAL WORK | % 60 | |
| total | % 100 | |
Workload and ECTS Credit Calculation
| Activities | Number of Activities | Preparation for the Activity | Spent for the Activity Itself | Completing the Activity Requirements | Workload | ||
| Midterms | 1 | 60 | 1 | 61 | |||
| Final | 1 | 80 | 1 | 81 | |||
| Total Workload | 142 | ||||||