NTHU EE6550 Machine Learning slides and my code solutions for spring semester 2017.

• The assignments are mainly required to implement the ML/DL algorithms from scratch without using high-level library like scikit-learn, all are coded with python/numpy.

• PAC Learning Rectangle
• Linear SVM: Linear Support Vector Machine for Binary Classification trained with Sequential Minimal Optimization.
• Kernel SVM and Adaboost
  • An Extension of HW2 linear SVM.
  • Adaboost classifer with the shallow decision tree (depth 1) for binary classification.
• Kernel SVR: Support Vector Regression, trained with Sequential Minimal Optimization.
• Neural Network: A simple neural network trained with SGD for regression.
• Final Project: A solution report of Kaggle competition - Quora Question Pairs. Using various deep neural nets and XGboost to identify duplicate questions.