理工学研究科Graduate School of Science and Engineering
COS500X3(計算科学 / Computational science 500)ニューラルネットワークの理論と応用Theory and Applications of Neural Networks
斉 欣Xin QI
授業コードなどClass code etc
学部・研究科Faculty/Graduate school | 理工学研究科Graduate School of Science and Engineering |
添付ファイル名Attached documents | |
年度Year | 2023 |
授業コードClass code | YB038 |
旧授業コードPrevious Class code | |
旧科目名Previous Class title | |
開講時期Term | 秋学期授業/Fall |
曜日・時限Day/Period | 月3/Mon.3 |
科目種別Class Type | |
キャンパスCampus | 小金井 |
教室名称Classroom name | 各学部・研究科等の時間割等で確認 |
配当年次Grade | |
単位数Credit(s) | 2 |
備考(履修条件等)Notes | |
実務経験のある教員による授業科目Class taught by instructors with practical experience | |
カテゴリーCategory | 応用情報工学専攻 |
すべて開くShow all
すべて閉じるHide All
Outline (in English)
The course covers both theoretical and practical aspects of neural networks, including their implementation.
By the end of the course, students will gain a solid understanding of neural network principles and will be able to apply them to solve some computer vision and signal processing problems.
In addition to classroom instruction, students are expected to devote about an hour to learning activities per class.
Grading will primarily be based on the quality of the project and presentation.
授業で使用する言語Default language used in class
英語 / English
授業の概要と目的(何を学ぶか)Outline and objectives
The course covers both theoretical and practical aspects of neural networks, including their implementation.
By the end of the course, students will gain a solid understanding of neural network principles and will be able to apply them to solve some computer vision and signal processing problems.
In addition to classroom instruction, students are expected to devote about an hour to learning activities per class.
Grading will primarily be based on the quality of the project and presentation.
到達目標Goal
There are 3 major goals.
1) Understand the basic principles of neural networks.
2) Command at least one training framework such as Tensorflow.
3) Can solve one computer vision or signal processing research problems by using neural networks.
この授業を履修することで学部等のディプロマポリシーに示されたどの能力を習得することができるか(該当授業科目と学位授与方針に明示された学習成果との関連)Which item of the diploma policy will be obtained by taking this class?
ディプロマポリシーのうち、「DP1」「DP2」「DP3」に関連
授業で使用する言語Default language used in class
英語 / English
授業の進め方と方法Method(s)(学期の途中で変更になる場合には、別途提示します。 /If the Method(s) is changed, we will announce the details of any changes. )
There are 12 classes for lectures and exercises, 2 classes for the presentation.
Classes are provided via face-to-face by default and could be online. Changes in the lecture plan due to this shift will be announced on the learning support system.
アクティブラーニング(グループディスカッション、ディベート等)の実施Active learning in class (Group discussion, Debate.etc.)
あり / Yes
フィールドワーク(学外での実習等)の実施Fieldwork in class
なし / No
授業計画Schedule
授業形態/methods of teaching:対面/face to face
※各回の授業形態は予定です。教員の指示に従ってください。
1[対面/face to face]:History of Neural Networks
This course introduces the the history and inspiration of neural networks.
2[対面/face to face]:Training the Network
This course explains some basic knowledge for training the network.
3[対面/face to face]:Improve the Learning I
This course explains the training techniques such as cost function determinations.
4[対面/face to face]:Improve the Learning II
This course explains the training techniques such as regularization methods.
5[対面/face to face]:Convolutional Neural Network
This course introduces the structure and benefit of convolutional neural networks.
6[対面/face to face]:Variants of Convolutional Neural Network
This course introduces several kinds of convolutions such as transposed convolution.
7[対面/face to face]:Popular Convolutional Neural Network Architectures
This course introduces some recent famous CNN architectures such as AlexNet.
8[対面/face to face]:Reducing Complexity of Convolutional Neural Network
This course introduces some simplified CNN such as 1x1 convolution.
9[対面/face to face]:Advanced Convolutional Neural Network
This course introduces some advanced CNN such as group convolution.
10[対面/face to face]:Recurrent Neural Networks
This course explains some RNNs such as Long-Short Term Memory (LSTM).
11[対面/face to face]:Variational Autoencoder and Generative Adversarial Network
This course explains the principles of VAE and GAN and their usage in image generation.
12[対面/face to face]:Transfer Learning
This course shows how to use transfer learning in different networks.
13[対面/face to face]:Final Presentation I
Students give presentation.
14[対面/face to face]:Final Presentation II
Students give presentation.
授業時間外の学習(準備学習・復習・宿題等)Work to be done outside of class (preparation, etc.)
For each class, students should prepare for 2 hours and review for 2 hours, a total of 4 hours.
For every 3-4 classes, there is an exercise.
テキスト(教科書)Textbooks
No textbook will be used.
参考書References
No references will be used.
成績評価の方法と基準Grading criteria
Exercise or Report: 40%
Class participation & Attendance: 10%
Final project presentation: 50%
学生の意見等からの気づきChanges following student comments
Not applicable.
学生が準備すべき機器他Equipment student needs to prepare
A laptop for in-class use.