CURRENT TOPICS in DEEP LEARNING

AIM

This course is an introduction to deep learning, a branch of machine learning concerned with the development and application of modern neural networks. We will cover a range of topics from basic neural networks, convolutional and recurrent network structures, deep unsupervised learning, and applications to problem domains like computer vision, image processing and natural language processing. The course will introduce training and optimization strategies in deep networks both for supervised and unsupervised learning tasks. In this course, AI is a co-pilot. Following strategies will be used: AI-Assisted Coding: Students are encouraged to use tools like GitHub Copilot or ChatGPT to boilerplate code, provided they can explain and document every line generated. Socratic AI Tutoring: Using LLMs to simplify complex concepts (e.g., "Explain Backpropagation like I'm 5"). AI-Generated Synthetic Data: Using AI to create edge-case datasets for testing model robustness. Red-Teaming AI: Critiquing AI-generated code for "hallucinations" or architectural inefficiencies.

CONTENT

This course contains; Introduction to Machine Learning and Neural Networks,Training Neural Networks,Convolutional Neural Networks (CNNs),Network Layers in CNNs,Deep Learning Hardware and Software,Deep Network Architectures,Deep Learning Strategies,Recurrent Neural Networks and LSTMs,Natural Language Processing with Deep Learning,Computer Vision and Deep Learning,Image processing and Deep Learning,Unsupervised Learning and Generative Modeling,Advanced Applications of Deep Learning,Project Presentations.

1. 1

   1. Design Convolutional Neural Networks for supervised and unsupervised learning, while using AI assistants to compare different architectural trade-offs.

   Taught by: Problem Solving Method, Project Based Learning Model, Simulation Technique, Experiential Learning, Lecture Method · Assessed by: Traditional Written Exam, Homework, Project Task

2. 2

   2. Analyze the effects of hyper-parameters on learning performance using AI-driven simulation tools to visualize Loss Functions.

   Taught by: Problem Solving Method, Project Based Learning Model, Simulation Technique, Experiential Learning, Lecture Method · Assessed by: Project Task

3. 3

   3. Apply advanced learning techniques for training deep networks and use LLMs to troubleshoot Gradient issues.

   Taught by: Problem Solving Method, Project Based Learning Model, Simulation Technique, Experiential Learning, Lecture Method · Assessed by: Traditional Written Exam, Homework, Project Task

4. 4

   4. Implement applications of deep networks in computer vision, image processing, and natural language processing through AI-collaborative coding.

   Taught by: Project Based Learning Model, Simulation Technique, Experiential Learning, Lecture Method · Assessed by: Homework, Project Task

5. 5

   5. Use current software and hardware tools for deep learning, including Generative AI platforms for code generation, documentation, and synthetic data augmentation.

   Taught by: Project Based Learning Model, Simulation Technique, Experiential Learning, Lecture Method · Assessed by: Homework, Project Task

WEEKLY PLAN

1. WEEK 1

   Introduction to Machine Learning and Neural Networks

2. WEEK 2

   Training Neural Networks

3. WEEK 3

   Convolutional Neural Networks (CNNs)

4. WEEK 4

   Network Layers in CNNs

5. WEEK 5

   Deep Learning Hardware and Software

6. WEEK 6

   Deep Network Architectures

7. WEEK 7

   Deep Learning Strategies

8. WEEK 8

   Recurrent Neural Networks and LSTMs

9. WEEK 9

   Natural Language Processing with Deep Learning

10. WEEK 10

    Computer Vision and Deep Learning

11. WEEK 11

    Image processing and Deep Learning

12. WEEK 12

    Unsupervised Learning and Generative Modeling

13. WEEK 13

    Advanced Applications of Deep Learning

14. WEEK 14

    Project Presentations

ASSESSMENT

• Rate of Midterm Exam to Success50%
• Rate of Final Exam to Success50%

WORKLOAD

READING

• Deep Learning, I. Goodfellow, Y. Bengio and A. Courville , MIT Press, http://www.deeplearningbook.org , 2016.
• Machine Learning, Andrew Ng,Intel® AI Academy Deep Learning 501 https://software.intel.com/en-us/ai-academy/students/kits/deep-learning-501

TEACHING STAFF

• Prof.Dr. Bahadır Kürşat GÜNTÜRKCOORDINATOR
• Assist.Prof. Ahmet KAPLAN