mnist_tutorial

A tutorial for MNIST handwritten digit classification using sklearn, PyTorch and Keras.

Code structure

• numpy_matplotlib_sklearn.ipynb: for numpy, matplotlib and sklearn.
• pytorch.ipynb: for pytorch.
• keras.ipynb: for keras.
• Reference solution: (not published yet)
  • numpy_matplotlib_sklearn_solution.ipynb
  • pytorch_solution.ipynb
  • keras_solution.ipynb

Requirements

Code tested on following environments, other version should also work:

• linux system (ubuntu 16.04)
• python 3.6.3
• numpy 1.13.3
• matplotlib 2.1.0
• sklearn 0.19.1
• pytorch 0.4.1
• keras 2.1.2

For students from SJTU

Please read HEAR.

Results of Experiments

We use pytorch to solve Q5.

Question	Model	Train_Acc	Test_Acc
Q1	LogisticRegression(defalt)	97.20%	87.90%
Q2	BernoulliNB(default)	81.63%	81.90%
Q3	LinearSVC(default)	97.68%	87.10%
Q4	LinearSVC(adjusted)	93.63%	89.10%
Q5	SimpleNet()	99.76%	99.10%