liigoqi/data620004-hw1

MLP for MNIST

Yifan Qi

• MNIST Dataset
• Download Trained Model

Parameter searching:

python3 paramSearch.py

The best combination of learning rate (0.1, 0.05, 0.01), dimension of hidden layer (64, 144, 256) and the parameter of L2 regularization (1, 0.1, 0.01) is found. The result of all combinations of parameters is saved in search_parameters.csv.

Train with the best hyper-parameters:

python3 train.py

Before training, data should be downloaded and put in data folder. After training, the trained model is saved as model.pkl.

Test:

python3 test.py

model.pkl is loaded and used to test. The model precisions of all digital numbers are printed.