E5770 教學大綱表

教學大綱表 (112學年度第2學期)

請遵守智慧財產權，勿使用非法影印教科書，避免觸法。

課程名稱 Course Title	(中文) 類神經系統及實驗 (英文) Artificial Neural Systems And Laboratory			開課單位 Departments	電機工程研究所
				課程代碼 Course No.	E5770
授課教師 Instructor	許超雲
學分數 Credit	3.0	必/選修 core required/optional	選修	開課年級 Level	研究所
先修科目或先備能力(Course Pre-requisites)：
課程概述與目標(Course Overview and Goals)：建立基本機器學習之基礎及未來研究之能力
教科書(Textbook)
參考教材(Reference)

課程大綱 Syllabus			學生學習目標 Learning Objectives	單元學習活動 Learning Activities	學習成效評量 Evaluation	備註 Notes
序 No.	單元主題 Unit topic	內容綱要 Content summary	學生學習目標 Learning Objectives	單元學習活動 Learning Activities	學習成效評量 Evaluation	備註 Notes
1	Neural network 簡介:	歷史及Perceptron model	了解類神經系統定義基礎	上機實習講授個案研究		Case study
2	Adline/madline model 及Widraw-Hoff Learning	基礎/理論課程授課及Tenserflow及keras入手	1. basic learning algorithms 2. tensorflow 及 Keras 入手	上機實習講授實作		第一個實驗之(a): 華氏對攝氏溫度之轉換 (a): 熟習 Anaconda
3	Adline/madline及Backpagation Learning Rule	1)Widrow learning 2)Multi-layer Perceptron 3) Backpagation Learning Rule	Neural network= architecture + learning rule	上機實習講授		第一個實驗之(a): 華氏對攝氏溫度之轉換: 程式試RUN及修改
4	introduction to backprogation	BP Architecture and learning Rule	Architcture and learning rule	上機實習演講		從acitvation function: signmiodal 到 Rectifier Linear Unit 到 L到RELU對學習的影響
5	Complexity of machine learning	Optimizer and Loss function for machine learning	The complexity of learning and related performance index	討論講授		第一個實驗之(b)): 華氏對攝氏溫度之轉換以MLP+BP方式實現
6	Introduction to Autoencoder(AE)	1.Autoencoder= Encoder + Decoder without Supervision 2. PCA Vs. Autoencoder 3. Laternal space	AE' s adavantage and Bottleneck	上機實習演講		實驗二以自編碼器(AE)實現MNIST資料庫手寫數字之資料降維
7	Variational Autoencoder (VAE)	1) 4 disadvantages of AE 2) VariationalAE 3) Crossentropy Vs. KL Diveregence 4) Random generator	Combined learnig algorithm of VAE: Reconstruction loss + KL divergence	上機實習講授
8	Midterm	Midterm	Midterm
9	comparison of AE., VAE. and PCA	Difference among AE, VAE and PCA	Difference among AE VAE and PCA; and their poential usage	討論講授個案研究		case study
10	Convolutional Neural network(CNN)	1. Simple sensor makes contribution to visual intellengence. 2. CNN　architecture 3. Examples demo.	Architecture and learning rule of CNN	上機實習講授個案研究		實驗三以變分自編碼器(VAE)實現 MNIST資料降維分析與產生測試
11	Feedback NN	1) RNN 2) LSTM 3)GRU	1) Learning through BPTT 2) Gradient Varnished	上機實習講授個案研究
12	Support vector machine	Margin/support vector	1) Margin/support vector 2) Linear programing	上機實習講授個案研究		實驗四以LSTM自動生成音樂
13	Generator and Adversarial Neural Network(GAN)	Architecture and learning rule of GAN	1) Loss function of Generator and that of Adverial network 2) Convergence? 3) Improvement Tech.	上機實習講授個案研究
14	reinforcement learning	reinforcement learning　（ＲＬ）	1) Basic RL　theorem and applications 2)Ｍarkov decision Processs 3)Deep RL	上機實習講授個案研究
15	Introduction to transformer	1) basics of Transformer 2) encoder decoder stucture 3) Chat-GPT	1) basics of Transformer 2) encoder decoder stucture 3) Chat-GPT	上機實習講授個案研究		期末實驗: A small version Chat-GPT
16	Final exam	final exam	final exam
17	彈性教學 Final project competion	A small version of Chat-GPT	練習以現有開源之組件, 發展一小型 Chat-GPT	實作		10%
18	彈性教學 Final project competion	presentation and demo of self-developed small-version Chat-GPT	presentation and demo of self-developed small-version Chat-GPT	實作心得發表		total 10%

教學要點概述：
1.自編教材 Handout by Instructor：
□ 1-1.簡報 Slids
□ 1-2.影音教材 Videos
□ 1-3.教具 Teaching Aids
□ 1-4.教科書 Textbook
□ 1-5.其他 Other
□ 2.自編評量工具/量表 Educational Assessment
□ 3.教科書作者提供 Textbook

成績考核 Performance Evaluation：

教學資源(Teaching Resources)：
□ 教材電子檔(Soft Copy of the Handout or the Textbook)
□ 課程網站(Website)
扣考規定：https://curri.ttu.edu.tw/p/412-1033-1254.php