人工智慧與序列式決策
AI in Sequential Decision Making
| 節 | 週二 | 週三 |
|---|---|---|
8 16:30–17:20 | 人工智慧與序列式決策 MB415（光復） 2 節連堂 | |
9 17:30–18:20 | 人工智慧與序列式決策 MB415（光復） |
* 根據陽明交大上課時間表所列
This course introduces artificial intelligence methodologies for sequential decision-making problems, where decisions must be made repeatedly over time and each action influences future system states and outcomes. The course emphasizes decision modeling, learning-based heuristics, and empirical evaluation, rather than low-level algorithmic engineering or neural network design. Reinforcement learning, approximate dynamic programming, hybrid heuristics, and emerging LLM-assisted decision frameworks are presented as black-box decision engines for solving complex optimization and control problems in manufacturing, logistics, energy systems, and other engineered systems.
Programming
無備註
All the materials can be downloaded from the online course registration system.
Participation (5%) Homework (25%) Midterm paper presentation (25%) Term project (45%)
Learning-Based Decision Methods
Reinforcement learning, deep reinforcement learning (as function approximation), adaptive metaheuristics, hybrid decision systems
- 講授：
- 18
- 示範：
- 6
Advanced Topics and Applications
Constraint-aware decision making, online and adaptive decisions, LLM-assisted decision frameworks, project-oriented discussions
- 講授：
- 6
- 示範：
- 6
Foundations of Sequential Decision Making
Sequential decision problems, decision modeling, dynamic programming intuition, approximate dynamic programming
- 講授：
- 18
- 示範：
- 6
| 週次 | 主題 |
|---|---|
| 第 1 週 | Course overview and motivation: sequential decision making 2026-09-08(二),2026-09-09(三) |
| 第 2 週 | Decision modeling: state, action, and reward design 2026-09-15(二),2026-09-16(三) |
| 第 3 週 | Dynamic programming (DP) intuition 2026-09-22(二),2026-09-23(三) |
| 第 4 週 | Approximate dynamic programming (ADP) 2026-09-29(二),2026-09-30(三) |
| 第 5 週 | Reinforcement learning (RL) for decision problems 2026-10-06(二),2026-10-07(三) |
| 第 6 週 | Deep reinforcement learning (DRL) as function approximation 2026-10-13(二),2026-10-14(三) |
| 第 7 週 | Adaptive metaheuristics and heuristic selection 2026-10-20(二),2026-10-21(三) |
| 第 8 週 | Midterm paper presentation 2026-10-27(二),2026-10-28(三) |
| 第 9 週 | Hybrid learning-based decision systems 2026-11-03(二),2026-11-04(三) |
| 第 10 週 | Constraint-aware sequential decision making 2026-11-10(二),2026-11-11(三) |
| 第 11 週 | Learning-augmented and online decisions 2026-11-17(二),2026-11-18(三) |
| 第 12 週 | Multi-stage and decentralized decision problems 2026-11-24(二),2026-11-25(三) |
| 第 13 週 | LLM-assisted decision-making frameworks 2026-12-01(二),2026-12-02(三) |
| 第 14 週 | Term project 2026-12-08(二),2026-12-09(三) |
| 第 15 週 | Term project 2026-12-15(二),2026-12-16(三) |
| 第 16 週 | Term project 2026-12-22(二),2026-12-23(三) |
The lecture is given based on handouts. Parts of the handouts are referred to the following books and articles: Bertsekas, D. P. Dynamic Programming and Optimal Control, Vol. I & II. Athena Scientific, 4th Edition, 2017. Powell, W. B. Approximate Dynamic Programming: Solving the Curses of Dimensionality. Wiley, 2nd Edition, 2011. Sutton, R. S., & Barto, A. G. Reinforcement Learning: An Introduction. MIT Press, 2nd Edition, 2018. Rao, A., & Jelvis, T. Foundations of Reinforcement Learning with Applications in Finance. Chapman & Hall/CRC Press, 2023. Supplementary Reading : Selected journal articles from International Journal of Production Research, Computers & Industrial Engineering, Applied Soft Computing, Robotics and Computer-Integrated Manufacturing, IEEE Transactions on Industrial Informatics, European Journal of Operational Research, and related journals (assigned during the semester).
- 地點
- MB501
- 時間
- Tue. 12:00-13:00 (e-mail contact in advance)
- 聯絡方式
- cclin321@nycu.edu.tw