Course Syllabus

Course Information：

Year-Semester Course Number Course Name Phase Credits Hours Required/Elective Instructor Class Student NO. Withdraw Note

Spring Semester
Year 2019 259787 Building Deep Learning Applications 1 3.0 3 ★ Kuan-Ting Lai 產業接軌課程(大)
4EN4A
4EN4B 18 4 電子大四、電子所.電資外生所.產業接軌合開

Year-Semester	Course Number	Course Name	Phase	Credits	Hours	Required/Elective	Instructor	Class	Student NO.	Withdraw	Note
Spring Semester Year 2019	259787	Building Deep Learning Applications	1	3.0	3	★	Kuan-Ting Lai	產業接軌課程(大) 4EN4A 4EN4B	18	4	電子大四、電子所.電資外生所.產業接軌合開

Course Plan：

Instructor Kuan-Ting Lai

Email ktlai@ntut.edu.tw

Last Updated 2019-01-10 13:11:52

Course Objective The goal of this course is to introduce latest deep learning research and teach students how to develop deep learning applications using Python library (TensorFlow & Keras). Additionally, we cooperate with Qualcomm, which will send engineers to demonstrate using their Neural Processing SDK. The content includes: • Number recognition of License plate and Doorplate • Image Classification for 1000 categories • Object Detection for 9000 objects • Convolutional Neural Networks (CNN) • Recurrent Neural Networks (RNN) • Long-Short Term Memory (LSTM) • Generative Adversarial Networks (GAN) • Embedded Deep Learning Algorithms • Deep Reinforcement Learning • Qualcomm Embedded AI SDK 課程主要為介紹學術界最新的深度學習研究成果，教導學員使用Python深度學習套件，讓學員能將深度學習技術應用於各個不同的領域。另外本課程與Qualcomm台灣高通合作, 高通將來介紹如何使用Snap Dragon平台上的嵌入式AI SDK. 課程內容包括： 1. 車牌與門牌數字辨識 2. 人臉辨識 3. 1000種影像分類 4. 辨識9000種物件 5. 自動駕駛模擬 6. 嵌入式深度學習演算法空虛, 寂寞, 覺得冷? 讓深度學習來幫助你: 7. 遞迴式神經網路(RNN)：找不到人聊天？訓練自己的聊天機器人 8. 對抗生成網路(GAN)：用GAN產生自己的夢中情人 9. 深度強化學習(DRL)：訓練AI跟你一起玩遊戲

Course Schedule 1. Introduction to Deep Learning 2. Applied Math for Machine Learning 3. Fundamentals of Machine Learning 4. TensorFlow & Keras + Word Embedding 5. Convolutional Neural Network (CNN) and Image Classification 6. Recurrent Neural Network (RNN) and Long Short-term Memory (LSTM) 7. Qualcomm SDK Training (1) 8. Training your own ChatBot 9. Object Detection a. YOLO 9000 – You Only Look Once b. Single-shot Multi-Box Detection 10. Deep Reinforcement Learning a. OpenAI GYM b. Playing Atari Games, Doom and Minecraft 11. Qualcomm SDK Training (2) 12. Generative Adversarial Networks a. Adversarial Attack for CNN 13. Qualcomm SDK Training (3) 14. Qualcomm SDK Training (4) 15. Virtual Environment for Visual Deep Learning 16. Student Project Demos 17. Visit to Qualcomm Taiwan’s Headquarters

Evaluation and grading policy Homework (50%), Final Project (50%)

Materials All right reserved. Do not reproduce without permission.

Foreign-language textbooks:Yes
Francois Chollet, “Deep Learning with Python, ” Manning, 2017 Ian Goodfellow, Yoshua Bengio, and Aaron Courville, “Deep Learning,” MIT Press, 2017 Latest publications on CVPR, NIPS, ICML, AAAI, ICLR

The access to curricular consultation

Note
　

Instructor	Kuan-Ting Lai
Email	ktlai@ntut.edu.tw
Last Updated	2019-01-10 13:11:52
Course Objective	The goal of this course is to introduce latest deep learning research and teach students how to develop deep learning applications using Python library (TensorFlow & Keras). Additionally, we cooperate with Qualcomm, which will send engineers to demonstrate using their Neural Processing SDK. The content includes: • Number recognition of License plate and Doorplate • Image Classification for 1000 categories • Object Detection for 9000 objects • Convolutional Neural Networks (CNN) • Recurrent Neural Networks (RNN) • Long-Short Term Memory (LSTM) • Generative Adversarial Networks (GAN) • Embedded Deep Learning Algorithms • Deep Reinforcement Learning • Qualcomm Embedded AI SDK 課程主要為介紹學術界最新的深度學習研究成果，教導學員使用Python深度學習套件，讓學員能將深度學習技術應用於各個不同的領域。另外本課程與Qualcomm台灣高通合作, 高通將來介紹如何使用Snap Dragon平台上的嵌入式AI SDK. 課程內容包括： 1. 車牌與門牌數字辨識 2. 人臉辨識 3. 1000種影像分類 4. 辨識9000種物件 5. 自動駕駛模擬 6. 嵌入式深度學習演算法空虛, 寂寞, 覺得冷? 讓深度學習來幫助你: 7. 遞迴式神經網路(RNN)：找不到人聊天？訓練自己的聊天機器人 8. 對抗生成網路(GAN)：用GAN產生自己的夢中情人 9. 深度強化學習(DRL)：訓練AI跟你一起玩遊戲
Course Schedule	1. Introduction to Deep Learning 2. Applied Math for Machine Learning 3. Fundamentals of Machine Learning 4. TensorFlow & Keras + Word Embedding 5. Convolutional Neural Network (CNN) and Image Classification 6. Recurrent Neural Network (RNN) and Long Short-term Memory (LSTM) 7. Qualcomm SDK Training (1) 8. Training your own ChatBot 9. Object Detection a. YOLO 9000 – You Only Look Once b. Single-shot Multi-Box Detection 10. Deep Reinforcement Learning a. OpenAI GYM b. Playing Atari Games, Doom and Minecraft 11. Qualcomm SDK Training (2) 12. Generative Adversarial Networks a. Adversarial Attack for CNN 13. Qualcomm SDK Training (3) 14. Qualcomm SDK Training (4) 15. Virtual Environment for Visual Deep Learning 16. Student Project Demos 17. Visit to Qualcomm Taiwan’s Headquarters
Evaluation and grading policy	Homework (50%), Final Project (50%)
Materials	All right reserved. Do not reproduce without permission. Foreign-language textbooks:Yes Francois Chollet, “Deep Learning with Python, ” Manning, 2017 Ian Goodfellow, Yoshua Bengio, and Aaron Courville, “Deep Learning,” MIT Press, 2017 Latest publications on CVPR, NIPS, ICML, AAAI, ICLR
The access to curricular consultation
Note