| Course Name | Code | Semester | T+U Hours | Credit | ECTS |
|---|---|---|---|---|---|
| Introduction To Deep Learning | ENF 547 | 0 | 3 + 0 | 3 | 6 |
| Precondition Courses | |
| Recommended Optional Courses | |
| Course Language | Turkish |
| Course Level | yuksek_lisans |
| Course Type | Optional |
| Course Coordinator | Doç.Dr. DEVRİM AKGÜN |
| Course Lecturers | |
| Course Assistants | |
| Course Category | Field Proper Education |
| Course Objective | To teach mathematical fundamentals about deep learning, to use open source libraries related to deep learning, to develop deep learning applications. |
| Course Content | Mathematical foundations, tensor operations, Graident descent, backpropagation, Keras deeplearning library and examples of usage, Models of machine learning, Convolution neural networks (convnets), feature extraction with pre-trained convnet , convnet visualization, deep learning with text data , recurrent neural networks, 1D convnets for array processing, Keras functional API, Keras functions, TensorBoard visualization tool, Generative deep learning, Contemporary issues |
| # | Course Learning Outcomes | Teaching Methods | Assessment Methods |
|---|---|---|---|
| 1 | Knows the basic topics about deep learning | Lecture, Question-Answer, Drilland Practice, Problem Solving, Project Based Learning, | Testing, Homework, Project / Design, |
| 2 | Knows 1D and 2D convnet-convolution layers. | Lecture, Question-Answer, Drilland Practice, Problem Solving, Project Based Learning, | Testing, Homework, Project / Design, |
| 3 | Uses open source libraries for deep learning | Lecture, Question-Answer, Drilland Practice, Problem Solving, Project Based Learning, | Testing, Homework, Project / Design, |
| 4 | Knows text processing, Embedding Layer, Simple RNN, LTSM and GRU layers | Lecture, Drilland Practice, Problem Solving, Project Based Learning, | Testing, Homework, Project / Design, |
| Week | Course Topics | Preliminary Preparation |
|---|---|---|
| 1 | Overview of artificial intelligence, machine learning and deep learning | Slides |
| 2 | Mathematical foundations, Gradient descent algorithms, loss functions, backpropagation. | Slides |
| 3 | Tensor operations with Python, Keras deep learning library, | Slides |
| 4 | Multi-label classification, Regression | Slides |
| 5 | Data preprocessing, overfitting prevention, weight regularization, dropout | Slides |
| 6 | 2-Dimensional Convolution (conv2D) Neural Networks (convnets), pooling | Slides |
| 7 | Augmentation of image data, pretrained networks | Slides |
| 8 | Fine tuning, convolution filters visualization | Slides |
| 9 | Deep learning with text data, Embedding layers | Slides |
| 10 | Recurrent neural networks | Slides |
| 11 | LSTM and GRU layers | Slides |
| 12 | Array processing with 1D convnets | Slides |
| 13 | Keras functional API, Multiple input or multiple output models | Slides |
| 14 | Generative deep learning | Slides |
| Resources | |
|---|---|
| Course Notes | <p>Weekly slides</p> |
| Course Resources | Goodfellow, Ian, et al. Deep learning. Vol. 1. Cambridge: MIT press, 2016. Chollet, Francois. Deep learning with python. Manning Publications Co., 2017. |
| Order | Program Outcomes | Level of Contribution | |||||
|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | |||
| 1 | X | ||||||
| 2 | X | ||||||
| 3 | X | ||||||
| 4 | X | ||||||
| 5 | X | ||||||
| 6 | X | ||||||
| 8 | X | ||||||
| 9 | X | ||||||
| Evaluation System | |
|---|---|
| Semester Studies | Contribution Rate |
| 1. Ara Sınav | 60 |
| 1. Proje / Tasarım | 20 |
| 1. Ödev | 10 |
| 2. Ödev | 10 |
| Total | 100 |
| 1. Yıl İçinin Başarıya | 50 |
| 1. Final | 50 |
| Total | 100 |
| ECTS - Workload Activity | Quantity | Time (Hours) | Total Workload (Hours) |
|---|---|---|---|
| Course Duration (Including the exam week: 16x Total course hours) | 16 | 3 | 48 |
| Hours for off-the-classroom study (Pre-study, practice) | 16 | 3 | 48 |
| Mid-terms | 1 | 15 | 15 |
| Assignment | 2 | 5 | 10 |
| Project / Design | 1 | 10 | 10 |
| Final examination | 1 | 20 | 20 |
| Total Workload | 151 | ||
| Total Workload / 25 (Hours) | 6.04 | ||
| dersAKTSKredisi | 6 | ||