Course Logistics
Course Logistics
Instructor: 陳縕儂 Yun-Nung (Vivian) Chen Time: Tuesday 234, 9:10-12:20
Location: 資104
Website: http://adl.miulab.tw
NTU COOL: https://cool.ntu.edu.tw/courses/175/
Email: [email protected]
◦To ensure timely response, email title should contain “[ADL2019]”
◦Do NOT send to our personal emails
Always check the up-to-date information from the website
NTU COOL
新的課程平台: NTU COOL
◦ 課程側錄上傳
◦ 作業手寫題直接上傳繳交 (還可以寫 code 呢!)
強大的助教團隊
◦ 論壇郵件回信
◦ TA Recitation
◦ TA Hours
Course Goal
The students are expected to understand
1. how deep learning works
2. how to frame tasks into learning problems
3. how to use toolkits to implement designed models, and 4. when and why specific deep learning techniques work for
specific problems
Pre-requisites
Course
◦Required: college-level calculus, linear algebra
◦Preferred: probability, statistics
Programming
◦proficiency in Python; all assignments will be in Python
◦GitHub; all assignments will be handed in via GitHub
◦Kaggle; all assignments will be submitted to Kaggle
(tutorial from Stanford) (tutorial) (website)
GPU resources are LIMITED, so please consider your available resources for taking this course
GitHub Student Pack
The student plan provides unlimited private repositories
◦make your assignments private before the due date
◦make them public afterwards
Grading Policy
4 Individual Assignment: 18% x 4 = 72%
◦GitHub code w/ README
◦ The score is given based on the ranking list
◦ Bonus points for outstanding performance
◦ Late policy: 25% off per day late afterwards
Final Group Project: 25%
◦GitHub code, Project document
◦ Bonus points for the outstanding work
Others: 5%
◦Write-up for the guest lecture/company visit
Understanding the difference between “collaboration” and “academic infraction”
Individual Assignments
A2. Word Representation A1. Dialogue Modeling
Final Group Project (2~5 persons)
Choose your preferred project topic
◦ Proposal (BONUS!): submit your proposal
◦ Get additional bonus if other groups choose the same the proposed topics
◦ Presentation
◦ Poster presentation
◦ Outstanding projects will be selected for company-sponsored awards/prizes
◦ Project Report & Code
◦ Wrap-up project report
◦ GitHub code submission w/ README
The project details will be announced later
How to Get the Registration Code?
Limit: ~100 students per course Requirements
◦ Available GPU Resources
◦ Programming skills
◦ Fill in the Google Form
Selection order if out of limit
◦ EECS Graduate = EECS (4-yr up) > EECS Others > Others
Tentative Schedule
Week Topic Assignment
1 2019/02/19 Course Logistics & Introduction
2 2019/02/26 Neural Network Basics &Guest Lecture by Dr. Yang
3 2019/03/05 Backpropagation + Word Representations A1 – Dialogue Modeling 4 2019/03/12 Recurrent / Recursive Neural Networks
5 2019/03/19 TA Recitation A2 – Word Embeddings
6 2019/03/26 Attention Mechanism 7 2019/04/02 Spring Break
8 2019/04/09 Word Embeddings + Contextual Embeddings A3 – Game Playing 9 2019/04/16 Company Workshop
10 2019/04/23 Convolutional Neural Networks
11 2019/04/30 Deep Reinforcement Learning A4 – Conditional Generation 12 2019/05/07 Deep Reinforcement Learning
13 2019/05/14 Break
14 2019/05/21 Generative Adversarial Networks 15 2019/05/28 Generative Adversarial Networks 16 2019/06/04 Break
17 2019/06/11 Unsupervised Learning 18 2019/06/18 Final Project Presentation
