課程名稱 |
深度學習應用於電腦視覺 Deep Learning in Computer Vision |
開課學期 |
110-2 |
授課對象 |
工學院 電腦輔助工程組 |
授課教師 |
吳日騰 |
課號 |
CIE5151 |
課程識別碼 |
521EU9310 |
班次 |
|
學分 |
3.0 |
全/半年 |
半年 |
必/選修 |
選修 |
上課時間 |
星期一6(13:20~14:10)星期四5,6(12:20~14:10) |
上課地點 |
普405普405 |
備註 |
本課程以英語授課。 限本系所學生(含輔系、雙修生) 總人數上限:50人 |
Ceiba 課程網頁 |
http://ceiba.ntu.edu.tw/1102CIE5151_ |
課程簡介影片 |
|
核心能力關聯 |
本課程尚未建立核心能力關連 |
課程大綱
|
為確保您我的權利,請尊重智慧財產權及不得非法影印
|
課程概述 |
This course introduces the fundamental theory/background knowledge of prevalent machine learning (ML) and computer vision (CV) algorithms. Relevant applications in the broad domain of the engineering community will be introduced to motivate the students. The first half of the semester will focus on the reasoning of artificial intelligence, several ML algorithms, model evaluation, deep learning (DL) and reinforcement learning. The rest of the semester will have emphasis on the reasoning of image processing, image feature extractions and pairing, as well as image-based sensing. After taking this course, students are expected to be equipped with basic knowledge and implementation skills to develop ML, DL or CV based approaches for applications in engineering. |
課程目標 |
Upon taking this course, students are anticipated to be well-prepared in the following items:
1. Understand the fundamental principles that support the ML/DL algorithms.
2. Be able to reasoning the performance of ML/DL models.
3. Be able to implement ML/DL algorithms.
4. Understand the fundamental principles that support the CV algorithms.
5. Understand the image representations of the world.
6. Be able to implement CV algorithms. |
課程要求 |
Prerequisites: Calculus, Computer Programming |
預期每週課後學習時數 |
|
Office Hours |
每週一 14:30~16:30 備註: Absence of the class will be allowed only if the student informed the instructor in advance. |
指定閱讀 |
No required textbook. |
參考書目 |
Several excellent online sources are:
1. A Course in Machine Learning, electronic source available at: http://ciml.info/
2. Christopher Bishop (2006), Pattern Recognition and Machine Learning, Springer
3. Goodfellow et. al (2016), Deep Learning, MIT Press, electronic source available at: https://www.deeplearningbook.org/
|
評量方式 (僅供參考) |
|
週次 |
日期 |
單元主題 |
Week 1 |
2/14, 2/17 |
Introduction to artificial intelligence, machine learning, and deep learning |
Week 2 |
2/21, 2/24 |
Data representations, k-nearest neighbor, decision tree |
Week 3 |
2/28, 3/3 |
2/28(beak), support vector machine |
Week 4 |
3/7, 3/10 |
Fully-connected neural network |
Week 5 |
3/14, 3/17 |
Fully-connected neural network (Cont.) |
Week 6 |
3/21, 3/24 |
Evaluation of machine learning models |
Week 7 |
3/28, 3/31 |
Convolutional neural network |
Week 8 |
4/4, 4/7 |
4/4(break), transfer learning and auto-encoder |
Week 9 |
4/11, 4/14 |
Generative adversarial network, Midterm I (4/14) |
Week 10 |
4/18, 4/21 |
Introduction to reinforcement learning and deep reinforcement learning |
Week 11 |
4/25, 4/28 |
Introduction to image basics and image-based sensing |
Week 12 |
5/2, 5/5 |
Image filtering |
Week 13 |
5/9, 5/12 |
Feature extraction and pairing |
Week 14 |
5/16, 5/19 |
Digital image correlation and image stitching |
Week 15 |
5/23, 5/26 |
World-image correspondence, Midterm II (5/26) |
Week 16 |
5/30, 6/2 |
3D reconstruction |
Week 17 |
6/6, 6/9 |
Introduction to probabilistic modeling (optional), Term project presentation |
Week 18 |
6/13, 6/16 |
Bayesian approaches (optional), Term project presentation |
|