Tamás Takács, Imre Molnár

16 min read

Deep Network Development

2024/25/2

My PhD colleague, Imre Molnár, and I will be compiling all lecture materials, exams, and practice notebooks for the upcoming termtime, along with a list of the topics covered. Our goal in making this content publicly available is to provide students with more opportunities to prepare for the course and to help them keep up with the course material as it is released weekly.

Please note that the materials may contain small mistakes, typos, or even implementation bugs. I would appreciate any notifications about these issues sent to my email.

Lecture and Practice Content

1. Lecture

Course introduction, course technical and administrative details.

Introduction

1. Practice

Python, NumPy and PyTorch Basics.

Python Introduction Python Fundamentals NumPy Introduction NumPy Fundamentals GPU PyTorch

2. Lecture

Linear Regression and MLPs.

Linear Regression

2. Practice

Life Expectanncy calculation with regression.

Autograd Linear Regression Data

3. Lecture

Convolution, Pooling and Convolutional Neural Networks.

CNNs

3. Practice

Image Classification using CNNs.

CIFAR-10 Image Classification Spectrogram Classification

4. Lecture

Transfer Learning and CNNs.

Transfer Learning

4. Practice

Transfer Learning in PyTorch.

Transfer Learning Quantization Techniques

5. Lecture

Introduction to Object Detection.

Object Detection

5. Practice

Object Detection with YOLO.

YOLO Object Detection Faster R-CNN Object Detection

6. Lecture

Image Segmentation with U-Net and Mask-RCNN.

Image Segmentation

6. Practice

Segmentation tools in PyTorch.

U-Net Brain MRI Segmentation YOLO Instance Segmentation YOLO Pose Estimation SAM 2

7. Lecture

Vanilla RNNs, LSTMs and GRUs.

RNNs

7. Practice

Sequental data processing in PyTorch.

Sentiment Analysis Time Series Data Machine Translation

8. Lecture

Introduction to the Transformer Architecture.

Transformer Architecture

8. Practice

The Attention Mechanism and Transformers

The Attention Mechanism Transformer from Scratch NMT with Transformers Hugging Face and Gradio

9. Lecture

Vision Transformers.

Vision Transformers

9. Practice

Vision Transformers in PyTorch.

The Vision Transformer Superpixel Tokenization

11. Lecture

Annotation Tools, Depth Estimation, Optical Flow.

Deep Learning Tools for CV part 1

11. Practice

Depth Estimation and Optical Flow in PyTorch.

Monocular Depth Estimation Optical Flow Image Inpainting

13. Lecture

Generative Modelling, Image Infilling and Neural Rendering.

Deep Learning Tools for CV part 2

14. Lecture

Department and Research Group Projects.

Department Projects

Assignments

The assessment is divided into three components. Students must complete two smaller, individual assignments that contribute to the practice grade and do not require oral defense. A larger, in-depth assignment, also part of the practice grade, must be defended orally, and the defense contributes to the lecture grade as well.

Additionally, there are four quizzes throughout the semester, each assessing understanding of lecture topics through both conceptual questions and minor calculations. These quizzes are administered exclusively via the university’s Canvas platform.

1. Homework

Image Colorization.

Assignment Details

2. Homework

Simplified Object Detection.

Assignment Details

1. Assignment

Image Captioning.

Assignment Details

Exams

Exams structure is currently in development.

Course Syllabus

Schedule

• Lecture: Fridays, 10:00 AM - 12:00 AM
• Practice: Fridays, 8:00 AM - 10:00 AM
• Location (for both): South Building, Lecture Hall 0-822 (Mogyoródi József terem)

Note:

• Hungarian: Déli Tömb 0-822 Mogyoródi József terem

Description

This course is designed to provide students with an in-depth exploration of Deep Learning, particularly focusing on Neural Network architectures. Throughout the semester, students will gain a comprehensive understanding of how Deep Neural Networks work, from the fundamental theory behind their design to practical implementation skills. The course primarily covers Supervised Deep Learning techniques and equips students with hands-on experience using PyTorch, a popular Deep Learning framework. By working through practices and assignments in PyTorch, students will learn to build, train, and optimize neural networks effectively.

The course also emphasizes ethical considerations in AI development, ensuring that students not only learn the technical aspects of Deep Learning but also understand its broader impact on society.

Grading

Your final grade is calculated using the weighted formula below:

Final Grade = 0.4 * Lecture (Quizzes and Assignment Defense) + 0.6 * Practice (Homeworks and Assignment)

• Pass required on a final written exam from the lecture and practice material (Pass/Fail Exam with Coding Pre-Exam)

Prerequisites

• Linear Algebra
• Probability Theory
• Programming Skills (for practices)

Tools and Frameworks

• Programming Language: Python
• Frameworks: PyTorch
• Libraries: NumPy, Matplotlib, torchvision, torchaudio
• Additional Tools: Goole Colab

Learning Objectives

• Understand the basics of Deep Learning
• Understand and implement Neural Network architectures
• Learn a popular Deep Learning framework (PyTorch)
• Be able to use open-source Neural Network software

Recommended Reading

1. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.
2. Paszke, A., et al. (2019). PyTorch: An Imperative Style, High-Performance Deep Learning Library.
3. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning.
4. Brownlee, J. (2019). Deep Learning for Computer Vision. Machine Learning Mastery.
5. Ethical AI Guidelines: Montreal Declaration for Responsible AI.

For any questions related to the course material, please message me or my colleague, Imre. For inquiries regarding access to the AI Lab, please contact Imre only or Kristian Fenech, the course owner. Happy learning!

3079 Words

01/19/2025 (Last updated: 2025-05-02 17:00:01 +0200)

f4442e5 @ 2025-05-02 17:00:01 +0200