ML_Tutorial

Sample Code (Python) and Data for Studying Machine Learning Algorithms

Prerequisite

1.Required Library: Numpy, Matplot, Scipy, Tensorflow

2.Knowledge of following topics

• Linear algebra
• Diffential formula (Matrices, Composite Function)
• Gradient Descent
• Vectorized Implementation with Numpy

Please refer to http://

1. Linear Regression

Input Data

• [lr_data1.txt] - Single Feature Example (X, Y) from Andrew Ng's ML class
• [lr_data2_multi.txt] - Multiple Feature Example (X1, X2, Y) from Andrew Ng's ML class
• [lr_data3.txt] - Single Freture Example. X0 is included in data file(1, X1, Y). from 'Machine Learning in Action'

Code

• [_test_linear_regression*.py] - Test Code for Linear Regression with input data
• [_test_linear_regression*_tf.py] - Test Code using TensorFlow for Linear Regression
• [linear_regression.py] - Functions for Linear regression
• [linear_regression_tf.py] - Functions for Linear regression (TensorFlow Version)
• [ml_utils.py] - Utility functions (i.e., featureNormalize(X))

2. Logistic Regression

Input Data

• [logistic_data1.txt] - Example from Andrew Ng's ML class (X1, X2, Y)
• [logistic_data2_reg.txt] - Example form Andrew Ng's ML class (+Polynomial and Regularization)
• [logistic_data3.txt] - Example from 'Machine Learning in Action' (X1, X2, Y)

Code

• [_test_logistic_regression.py] - Test Code for Logistic Regression with input data 1,3
• [_test_logistic_regression2.py] - Test Code for Logistic Regression with input data 2
• [_test_logistic_regression_tf.py] - Test Code using TensorFlow with input data 1,3
• [logistic_regression.py] - Functions for Logistic regression
• [logistic_regression_tf.py] - Functions for Logistic regression (TensorFlow Version)
• [ml_utils.py] - Utility functions (i.e., mapFeature(X): Polynomial Extension)

3. Softmax Regression

Input Data

• [mnist/*] - data file from MNIST

Code

• [_test_softmax.py] - Test code for Softmax Regression with Mnist Data
• [_test_softmax_tf.py] - Test code for Tensorflow Softmax Regression with Mnist Data
• [softmax.py] - Functions for Softmax Regression (Ref. UFLDL Tutorial)
• [softmax_tf.py] - Functions for Softmax Regression (TensorFlow version)
• [data_utils.py] - Utility Function for data handlng (load MNIST data)
• [ml_utils.py] - Utility Function (onehotTransform(X)): Categoricla Data to one-hot)

Reference

• Stanford Machine Learning Class in Coursera
• UFLDL Tutorial
• Tensor Flow