人工智慧

臺北科技大學 114-2 學期課程，課號 359540。

課程大綱

人類在從事數學演算、推理交談、觀察等各類動作時，均展現出 "智慧"的現象。人工智慧即是利用電腦結合基本理論在有限度的範圍內從事上述各項工作，其中涵蓋了－自然語言了解、機器人學、電腦視覺、類神經網路、專家系統、自動化定理證明等，而本課程則主要在於提供必要之基礎，主要內容包括：1.搜尋法則，2. 命題與敘述邏輯，3. 知識系統與知識表達， 4. 馬可夫鍊及其應用，5.最佳化技術，6. 機器學習， 7.增強式學習，8. 其他相關技術。 Artificial intelligence seeks to introduce the elements of intelligence into programs and systems in application domain. Topics will include 1. Searching methods 2. Propositional and Predicate logic 3. Knowledge-based systems and representation of knowledge 4. Markov chain and its applications, 5. Optimization techniques, 6.Machine learning, 7. Reinforcement learning, 8. Related topics.

課程進度

Week 1 Class announcement Introduction to artificial intelligence Week 2 Knowledge representation Semantic network Propositional logic & first-order logic Expert system Frames Problem solving & search BFS, DFS, UCS Week 3 Problem solving & search (Cont’d) Heuristics Search and A* Game trees Minimax α-β pruning Dynamic programming Week 4 Constraint satisfaction problems (CSP) Overview of optimalization Continuous & discrete functions Week 5: Continuous optimalization Gradient descent Lagrange multiplier Discrete functions Linear programming and interior-point method Branch-and-bound method Dynamic programming (again) Week 6 Metaheuristic optimization algorithms Genetic algorithm Simulated annealing Others Intro to probability Week 7 Markov chain Monte Carlo (MC) methods Computing pi via MC Week 8 MCMC (Markov Chain Monte Carlo) & Metropolis–Hastings algorithm Computing pi via MCMC Gibbs sampling Application in topic modeling (LDA) with Gibbs sampling Bayesian Networks & inference Week 9 MT Week 10 MT sol Hidden Markov chain (HMM) Week 11 HMM (cont’d) Application of HMM Week 12 Quick review of feedforward neural networks CNN and deep learning Week 13 Markov decision process (MDP) Introduction to reinforcement learning Policy function and value function Week 14 Introduction to reinforcement learning (Cont’d) Q learning DQN Policy-based RL REINFORCE Week 15 Policy-based RL (Cont’d) A3C PPO Week 16 Spare time if progress lags Additional topics (if time permits) Monte Carlo Tree Search (MCTS) Large-language models (LLMs) Week 17 Final exam Week 18 Presentation (Morning 10 -12, Internet)

評量方式

MT 50 % Final 50 % Term Project (Bonus) 10 %

教材

【遵守智慧財產權觀念，請使用正版教科書，不得使用非法影印教科書】 使用外文原文書：否 Lecture noted available.