- Implemented digit classification of MNIST images by Logistic Regression, Multilayer Perceptron Network and Convolutional Neural Network using TensorFlow Library.
- Tuned the hyper parameters - used different optimizers, varied learning rate, varied number of neurons and number of layers to improve the efficiency. Compared the efficiency for all three methods.
- Verified the ‘No Lunch Theorem’ by testing the trained model on USPS dataset The Report contains detailed results.
prachip23 / classification-on-mnist Goto Github PK
View Code? Open in Web Editor NEWImplementation and evaluation of classification algorithms- logistic regression, single hidden layer neural network, convolutional neural network on MNIST dataset