[AI7042-00] AI Reverse Engineering (2026 Fall)

Instructor: Prof. Sung-Ho Bae 

Mon. 15:00 - 17:30 (EE Bldg. 309)

Class objectives 

• • Students will reverse-engineer representative world-model systems by tracing their representations, objectives, training data, rollout mechanisms, planning modules, and failure modes. 

Midterm Exam (40%)

• • Covering all the lectures dealt with, taking for 60 min

  • Closed book exam

Final Project (40%)

• • Performing research on emerging issues in world models 

  • Preparing for 10 min presentation which effectively summarizes your work 

Homework (10%)

• • Homework will be distributed in every lecture and should be submitted to e-campus in one week (late submission is not allowed) 

Attendance (10%)

• • If the attendance score becomes 0, F grade will be assigned, (-2 for absence, -0.5 for tardy, it will be abscence if arrived after 1 hour)

  • F-grading will be given for 5 times abscence 

Ethical issues (judged by the campus law)

• • Cheating

References

• • Reinforcement learning: 

  • Transformers:

  • Generative models: 

    • https://diffusion.csail.mit.edu/ 

    • https://deepgenerativemodels.github.io/ 

    • The principles of diffusion models

  • World models

Week 01

• • 09/07 (Mon): Reinforcement Learning (1/4)

Week 02

• • 09/14(Mon): Reinforcement Learning (2/4)

Week 03

• • 09/21 (Mon): Reinforcement Learning (3/4)

Week 04

• • 09/28 (Mon):Reinforcement Learning (4/4)

Week 05

• • 10/05 (Mon): RNN and Transformer (1/3)

Week 06

• • 10/12 (Mon): RNN and Transformer (2/3)

Week 07

• 10/19 (Mon):  RNN and Transformer (3/3)

Week 08

• • 10/26 (Mon): Midterm Exam

Week 09

• 11/02 (Mon): Generative Models (1/2) 

Week 10

• • 11/09 (Mon): Generative Models (2/2)  

Week 11

• • 11/16 (Mon):  World Models (1/3)

Week 12

• • 11/23 (Mon): World Models (2/3)

Week 13

• • 11/30 (Mon):  World Models (3/3)

Week 14

• • 12/07 (Mon): Final Project Presentations (1/2)

Week 15

• 12/14 (Mon):  Final Project Presentations (2/2)