krz-max / introduction-to-machine-learning

[NYCU 2022 Fall] Introduction to Machine Learning

[NYCU 2022 Fall] Introduction to Machine Learning

Instructor: Prof. Yen-Yu Lin

• You could download the notebooks & data, put them all in the same folder and run all on your local environment or GoogleColab
• You need to unzip the data first
• You might need to download some package if you don't have some of them.

HW1: Logistic regression and Linear Regression using Gradient Descent

Goals

In this coding assignment, you need to implement linear regression by using only NumPy, then train your implemented model using Gradient Descent by the provided dataset and test the performance with testing data.

Find the questions at here

Output

• LinearRegression MSE is 54.2685 Intercept is [-0.4061] and weights are [50.7277]
• LogisticRegression Cross Entropy Error is 0.1854 Intercept is [4.5803] and weights are [1.5370]

Method	Learning Curve	Result
Logistic Regression
Linear Regression

HW2: Linear Discriminant Analysis

Goals

In hw2, you need to implement Fisher’s linear discriminant by using only numpy, then train your implemented model by the provided dataset and test the performance with testing data

Please note that only NUMPY can be used to implement your model, you will get no points by simply calling sklearn.discriminant_analysis.LinearDiscriminantAnalysis

Find the questions at https://docs.google.com/document/d/1T7JLWuDtzOgEQ_OPSgsSiQdp5pd-nS5bKTdU3RR48Z4/edit?usp=sharing

Projection Result

• From HW3 to HW5, see the result in the report.pdf

HW3: Decision Tree, AdaBoost and Random Forest

Goals

In this coding assignment, you need to implement the Decision Tree, AdaBoost and Random Forest algorithm by using only NumPy, then train your implemented model by the provided dataset and test the performance with testing data. Find the sample code and data on the GitHub page.

Please note that only NumPy can be used to implement your model, you will get no points by simply calling sklearn.tree.DecsionTreeClassifier.

Find the questions at this document

HW4: Decision Tree, AdaBoost and Random Forest

Goals

In this coding assignment, you need to implement the Cross-validation and grid search by using only NumPy, then train the SVM model from scikit-learn by the provided dataset and test the performance with testing data.

Please note that only NumPy can be used to implement cross-validation and grid search, you will get no points by simply calling sklearn.model_selection.GridSearchCV

Find the questions at this document

HW5: Simple Captcha Hacker using Neural Network

Goals

Implement the deep neural network by any deep learning frameworks, e.g., Pytorch, TensorFlow and Keras, and then train DNN model on the provided dataset Find the Kaggle page here

Example

• Train a model to predict all the digits in the image
  • Task 1: Single character in the image
  • Task 2: Two characters in the image (order matters)
  • Task 3: Four characters in the image (order matters)
• Download the Data Here

Final Project: Product Failure Prediction

• Download the Data Here
• For more information, read README.md in the folder Final_Project