purelyvivid/DeepLearning-MachineLearning_tutorial

For Deep Learning/Machine Learning Primer, FAREASTONE employee training

Deep Learning/Machine Learning Tutorial

• Objective

  • Simple example for DL/ML primer

• Toolkit

• Python 2.7 or 3.5

• Numpy version >= 1.13.1 Tensorflow version >= 1.0.1 Keras version >= 2.0.6 Sklearn version >= 0.18.2

  ​

Homework_1 ANN

• hw1_ANN_answer.ipynb

• Dataset: breast cancer wisconsin dataset (classification)

  Item	Description
  Classes	2
  Samples per class	212(M),357(B)
  Samples total	569
  Dimensionality	30
  Features	real, positive

• Model: Affine Neural Network(ANN) (fully-connected)

  

  ​

Homework_2 Linear Regression

• hw2_LinearRegression_1.ipynb
• Dataset: boston house-prices dataset (regression)

Item	Description
Samples total	506
Dimensionality	13
Features	real, positive
Targets	real 5. - 50.

• Model: **Linear Regression **

  

Homework_3 Logistic Regression

• hw3_LogisticRegression_1.ipynb

• Dataset: breast cancer wisconsin dataset (classification)

  Item	Description
  Classes	2
  Samples per class	212(M),357(B)
  Samples total	569
  Dimensionality	30
  Features	real, positive

• Model: Logistic Regression

Homework_4 SVM

• hw4_SVM.ipynb

• Dataset: Olivetti faces data-set from AT&T

  • Total 400 images, 64*64 pixels, 40 classes(40 people), 10 faces with various emotions each person

• Model: SVM (support vector machine)

  

Appendix_4 Kernel Model

• kernel _model_rbf.ipynb

• LinearRegression with kernel function => kernel model

  ​

Homework_5 Decision Tree

• Dataset: iris dataset (classification)

  Item	Description
  Classes	3
  Samples per class	50
  Samples total	150
  Dimensionality	4
  Features	real, positive

• Model: Decision Tree

  

Appendix_6 AdaBoost Classifier Implementation (without sklearn)

• AdaBoost.ipynb

• Dataset: breast cancer wisconsin dataset (classification)

  Item	Description
  Classes	2
  Samples per class	212(M),357(B)
  Samples total	569
  Dimensionality	30
  Features	real, positive

• Model: AdaBoost

  

Appendix_7 kMeans Implementation (without sklearn)

• kMeans.ipynb

• kMeans_validity-index.ipynb (More completed, with clustering validity index)

• Dataset: iris dataset (classification)

  Item	Description
  Classes	3
  Samples per class	50
  Samples total	150
  Dimensionality	4
  Features	real, positive

• Model: kMeans