Skip to content

This notebook demonstrate the use of PyTorch to create a Multi-Layer Perceptron for Image Classification on Fashion Mnist Dataset.

Notifications You must be signed in to change notification settings

SJ-Ray/FMNIST_MLP_PTH_Classifier

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 

Repository files navigation

Fashion Mnist Multi-Layer Perceptron Classifier

This notebook demonstrate the use of PyTorch to create a Multi-Layer Perceptron for Image Classification on Fashion Mnist Dataset.

I have used Fashion Mnist Dataset which contains:

  • 60K training datapoint.
  • 10K test datapoints.
  • 10 Categories.
Label Description
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot

The model used is a Pytorch Sequential model with the following architecture.

Sequential(
    (0): Linear(in_features=784, out_features=128, bias=True)
    (1): ReLU()
    (2): Dropout(p=0.2)
    (3): Linear(in_features=128, out_features=64, bias=True)
    (4): ReLU()
    (5): Dropout(p=0.2)
    (6): Linear(in_features=64, out_features=10, bias=True)
    (7): LogSoftmax()
  )

Classification report on the test data set:

Label f1-score precision recall support
0 T-shirt/top 0.800395 0.821501 0.780347 519.0
1 Trouser 0.966732 0.951830 0.982107 503.0
2 Pullover 0.743383 0.674322 0.828205 390.0
3 Dress 0.830612 0.814000 0.847917 480.0
4 Coat 0.751342 0.876827 0.657277 639.0
5 Sandal 0.931206 0.906796 0.956967 488.0
6 Shirt 0.576375 0.546332 0.609914 464.0
7 Sneaker 0.921105 0.934000 0.908560 514.0
8 Bag 0.959402 0.947257 0.971861 462.0
9 Ankle boot 0.945489 0.961759 0.929760 541.0
micro avg 0.843600 0.843600 0.843600 5000.0
macro avg 0.842604 0.843462 0.847292 5000.0
weighted avg 0.844092 0.850633 0.843600 5000.0

About

This notebook demonstrate the use of PyTorch to create a Multi-Layer Perceptron for Image Classification on Fashion Mnist Dataset.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published