/Different-CNN-Architectures-on-MNIST-dataset-

Experimented with different architectures and kernels on MNIST dataset using Convolutional Neural Networks.

Primary LanguageJupyter Notebook

Different-CNN-Architectures-on-MNIST-dataset-

Screenshot (69)

1] Kernel (5,5) performs better than other two kernels in my architecture.
2] We got max Test Accuracy of 99.44 when we apply padding+batchnormalization + dropout on kernel (5,5) with 3 LSTM layers and on kernel(7,7) with 2 LSTM layers
3]Most of test Accuracy is greater than 99.30