.. | ||
cnn.go | ||
linear.go | ||
main.go | ||
nn.go | ||
README.md |
Linear Regression, NN, and CNN on MNIST dataset
MNIST
-
MNIST files can be obtained from this source and put in
data/mnist
from root folder of this project. -
Load MNIST data using helper function at
vision
sub-package
Linear Regression
-
Run with
go clean -cache -testcache && go run . -model="linear"
-
Accuracy should be about 91.68%.
Neural Network (NN)
-
Run with
go clean -cache -testcache && go run . -model="nn"
-
Accuracy should be about 94%.
Convolutional Neural Network (CNN)
-
Run with
go clean -cache -testcache && go run . -model="cnn"
-
Accuracy should be about 99.3%.