Class Info:

• Time: Class1 (24610): Tue 9:00-12:00, Class2 (23504): Thr 9:00-12:00
• Location: Cluster Bd (학연산클러스터), Room 506

Textbook: Python Machine Learning, 2nd Ed, Raschka & Mirjalili, PACKT Publishing

Grading: (Attendance, Team-projects, Midterm, Final) = (10%, 50%, 20%, 20%)

• Grading Rublic

Links

• PBL Center: http://pbl.hanyang.ac.kr
• TA & Q/A:
  • Cluster Bd. Room 620 (AI LAB) / nomar0107@gmail.com,

Exams

• Midterm
  • Oct 23, 7~8:30 pm
  • Location: Cluster Bd.
    • Room 506: Tuesday class
    • Room 507: Thurs class
    • There will be no lecture on Tue/Thurs in this week (시험이 있는 주에는 수업이 없습니다).
  • Important Topics 2019 [link]
• Final
  • Dec 11 (Wed) 7:00 ~ 8:30 pm
  • Important Topics (final) [link]

Schedule

• Week 1: Introduction to AI [note]
• Week 2~4: Basic methods in ML (Logistic regression, Neural Networks, SVM) [note1, note2]
• Week 5~7 (PBL Case Study #1) (How to read MNIST files)
  • Teams (class1 , class2)
  • Notes
    • Preprocessing [note]
    • Model selection [note]
    • Cross validation & Grid Search summary [note]
  • Main goal: model selection
    • F1-score
    • Hyperparameters
      • Logistic regression: lambda, ||w||_2^2, ||w||_1
      • SVM: C
    • To save time, you can use the test (10k) set for training, and the training set (60k) for testing.
  • Team presentations (send presentation files to nomar0107@gmail.com)
  • Evaluations: (class1, class2) (You must login with Google account)
• Week 8: Midterm
• Week 9: Gradient-Based Learning [pdf]
• Week 10~12: (PBL Case Study #2)
  • Leaderboard (class1, class2)
  • Team composition (class1, class2)
  • Team & Member evaluation (class1, class2)
  • Notes
    • SGD Algorithm for solving SVM [note]
    • Unsupervised Learning (PCA)
    • Unsupervised Learning: LDA, Kernel PCA [note]
  • Code submission (class 1: by Nov 17, class 2: by Nov 26 midnight)
    • TA : nomar0107@gmail.com
    • Python code: team#.py <training_data> <test data>
    • Additional MNIST data
      • New 1k [images, labels]
      • D1+D2+New 1k data [images, labels]
        • We will use this as the new <training_data> for the next evaluation
    • Fix hyperparameters in your code.
    • Output labels per line, as explained in PBL 2 problem description.
• Week 13~16: PBL #3
  • Leaderboard (class1, class2)
  • Team & Member evaluation (class1, class2)
  • Guideline(PPT)
  • Code submission (class 1: by Dec 15, class 2: by Dec 17 midnight)
  • Clustering, Decison tree, Ensemble Methods
• Week 15 (Dec 11, Wed): Final Exam
  • Dec 11 (Wed) 7:00 ~ 8:30 pm