1] As dropout is increased from 0.2 to 0.5 with same batch size and architecture , number of epoch needed starts increasing
2] After reaching the particular epoch the model overfits.
3] When we dont apply BatchNormalization, no. of epoch becomes 4 after which the model starts overfitting.
4] For 5 Hidden Layers with dropout = 0.5 , epoch reached maximum which is 50.
5] All the test Accuracy is in the range of (98 - 98.5) %
6] For models without BatchNormalization we got less test accuracy of approx 97 %
-
Notifications
You must be signed in to change notification settings - Fork 0
sahildigikar15/MLP-Architetures-on-MNIST-dataset
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
Experimented with different architectures on MNIST dataset using MLPs with different dropouts.
Topics
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published