Spring 2024
Introduction to Deep Learning (딥러닝개론) - Graduate Course
Instructor: Dr. Muhammad Syafrudin
Course information
This course will introduce deep learning using Python and Keras, and is loaded with insights. You’ll learn practical techniques that can be applied to the real world, and important theory for perfecting neural networks. As you progress through each week, you’ll build your understanding through explanations and practical examples. You are expected to acquire the skills you need to start developing deep learning applications.
Lecture
Monday 화15:00-18:00 KST / Daeyang AI Center 센535
Office hours
Appointment by email (Office at AI Center 502호)
Prerequisites
Prior knowledge of Calculus, Linear Algebra, and Probability will be advantageous
Notes
Course material can be downloaded in ecampus and please be aware,
that we will not publicly release the homework assignments this year.
Please note that:
-All lectures are in English.
-Weekly lecture topics may be adjusted or changed without prior notice depending on the understanding level during the class.
1. Presentation:
Your presentation will consist of a 25-30 minute talk followed by a 10-15 minute Q&A session.
** Since most of the course material is based on research papers, active class participation is crucial for this semester. Therefore, there are a couple of things you need to do before and during the class.
2. Once you have selected your paper, please inform the class by dropping a message in ecampus, stating "OOO presents OOO paper." Each student should choose their paper at least two weeks before their scheduled presentation.
3. For the paper you have selected, you should prepare your presentation. Additionally, you are required to write a homework assignment called "paper critique" for ALL the other papers selected by your classmates. This assignment involves creating a 1-2 page summary for each paper chosen by your peers. The paper critique should include the following content:
- 3 main points of the reading: What is the paper about? What problem does it address? What solution does it propose?
- 3 strengths of the reading: What is the main novelty of the paper? What is its impact? If you were to write such a paper, what aspects would you consider?
- 3 potential improvements for the reading: Are there any weaknesses in the paper? How can it be extended and improved?
- 3 questions you have about the paper.
4. You must submit a hard copy of the paper critique before the class. Additionally, you need to upload your presentation file and paper critique on ecampus prior to your presentation.
5. Each presentation will be followed by a 10-minute Q&A session. During the Q&A session, all students should actively engage by asking questions or discussing the paper. This is mandatory for all presentations.
6. If you have a suggestion for a paper that is related to our class, you can propose it with prior approval from me.
Schedule
| 주차(Week) | 강의내용(Class Topic & Contents) | 강의활동유형(Class Type) |
|---|---|---|
| 1 | Course overview (3/4) | Lecture and discussion |
| 2 | What is deep learning?, The mathematical building blocks of neural networks (3/11) | Lecture and discussion |
| 3 | Introduction to Keras and TensorFlow, Getting started with neural networks: Classification and regression (3/18) | Lecture and discussion |
| 4 | Fundamentals of machine learning, The universal workflow of machine learning (3/25) | Lecture and discussion |
| 5 | Working with Keras: A deep dive (4/1) | Lecture and discussion |
| 6 | Introduction to deep learning for computer vision (4/8) | Lecture and discussion |
| 7 | Deep learning for timeseries (4/15) | Lecture and discussion |
| 8 | Midterm exam (4/15) | 시험 (Exam) |
| 9 | Deep learning for text (4/29) | Lecture and discussion |
| 10 | No class -- Public holiday (5/6) | No class |
| 11 | Paper with code presentation: ELSTOHY M. A. G. A., GAZA HAIFA (5/13) | Student presentation and discussion |
| 12 | Paper with code presentation: SRIWIJAYA Y. R., JORDAN D. J.(5/20) | Student presentation and discussion |
| 13 | Paper with code presentation: LE T. L., ABBAS REZK (5/27) | Student presentation and discussion |
| 14 | Paper with code presentation: ABDALLA AHMED, ABDULKHALEK ALFAKIH (6/3) | Student presentation and discussion |
| 15 | Paper with code presentation: AHMED S. A. E. M., SAID SALEM (6/10) | Student presentation and discussion |
| 16 | Paper with code presentation: 우수한 (6/17) | Student presentation and discussion |
Grading
The final grade will be calculated using the following weights:
| # | Final Grade Weight |
|---|---|
| Attendance | 15% |
| Assignment(quiz/homework/weekly assignment) & presentations | 45% |
| Mid exam | 40% |
| Total | 100% |
Assignment
There might be one or two student presentations, and in addition, paper critiques for other students' presentations will be assigned and evaluated.
Submitting an assignment
Instructions for turning in assignments will be posted when the semester starts (in ecampus).
Getting help
For questions about homework, course content, installation, and after you have tried to troubleshoot yourselves, the process to get help is: Post the question in ecampus/group chat and hopefully your peers will answer. Note that in ecampus questions are visible to everyone. For private matters send an email to helpline: udin [at] sju [dot] ac [dot] kr.
Course Policies
Collaboration policy
We encourage you to talk and discuss the assignments with your fellow students (and on ecampus), but you are not allowed to look at any other students assignment or code outside of your pair. Discussion is encouraged, copying is not allowed.
Late day policy
Homework is due before each class. Late submission are not allowed.
Communication to students
Class announcements will be through ecampus. All homework and quizzes will be posted in ecampus. Also all feedback forms. Important note: make sure you have your settings set so you can receive emails from ecampus.
Academic honesty
We give a strong emphasis to Academic Honesty. As a student your best guidelines are to be reasonable and fair. We encourage teamwork for problem sets, but you should not split the homework and you should work on all the problems together.
Presented Papers
- Futrega, M., Milesi, A., Marcinkiewicz, M., & Ribalta, P. (2022). Optimized U-Net for Brain Tumor Segmentation. In A. Crimi & S. Bakas (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries (Vol. 12963, pp. 15–29). Springer International Publishing. https://doi.org/10.1007/978-3-031-09002-8_2
- Balakrishnan, G., Zhao, A., Sabuncu, M. R., Guttag, J., & Dalca, A. V. (2019). VoxelMorph: A Learning Framework for Deformable Medical Image Registration. IEEE Transactions on Medical Imaging, 38(8), 1788–1800. https://doi.org/10.1109/TMI.2019.2897538
- Ledig, C., Theis, L., Huszar, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z., & Shi, W. (2017). Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (arXiv:1609.04802). arXiv. http://arxiv.org/abs/1609.04802
- Al-Fakih, A., Shazly, A., Mohammed, A., Elbushnaq, M., Ryu, K., Gu, Y. H., Al-masni, M. A., & Makary, M. M. (2024). FLAIR MRI sequence synthesis using squeeze attention generative model for reliable brain tumor segmentation. Alexandria Engineering Journal, 99, 108–123. https://doi.org/10.1016/j.aej.2024.05.008
- Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You Only Look Once: Unified, Real-Time Object Detection (arXiv:1506.02640). arXiv. http://arxiv.org/abs/1506.02640
- Wang, C.-Y., Yeh, I.-H., & Liao, H.-Y. M. (2024). YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information (arXiv:2402.13616). arXiv. http://arxiv.org/abs/2402.13616
- Liu, L., Du, B., Xu, J., Xia, Y., & Tong, H. (2022). Joint Knowledge Graph Completion and Question Answering. Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 1098–1108. https://doi.org/10.1145/3534678.3539289
- Nicodemus, J., Kneifl, J., Fehr, J., & Unger, B. (2022). Physics-informed Neural Networks-based Model Predictive Control for Multi-link Manipulators. IFAC-PapersOnLine, 55(20), 331–336. https://doi.org/10.1016/j.ifacol.2022.09.117
- Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J. M., & Luo, P. (2021). SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers (arXiv:2105.15203). arXiv. http://arxiv.org/abs/2105.15203
- Ren, S., He, K., Girshick, R., & Sun, J. (2016). Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (arXiv:1506.01497). arXiv. http://arxiv.org/abs/1506.01497
- Bu, M., Feng, T., & Lu, G. (2023). Prediction on local structure and properties of LiCl-KCl-AlCl3 ternary molten salt with deep learning potential. Journal of Molecular Liquids, 375, 120689. https://doi.org/10.1016/j.molliq.2022.120689