To fine tune moderately deep CNN so that it fits well to CIFAR-10 dataset AIM:To fine tune moderately deep CNN so that it fits well to CIFAR-10 dataset
Base Architecture of the model: The neural network has 2 convolution layers and 2 fully connected layers, followed by a softmax layer. The first convolution layer has 64 filters each of size 55. It is followed by a max pooling layer of size 22. The second convolution layer also has 64 filters each of size 5*5. This layer is also followed by a max pool layer. The size of the first fully connected layer is 256 and followed by another fully connected layer of size 128. Various parameters of the base model were varied to fine tune the CNN so that it fits the CIFAR-10 dataset well.
Please refer to the file AMP Assignment-1 Report.pdf for the different model variants that we have experimented with.
Collaborators: Meghana Kotagiri(meghanakotagiri) Neha Gaddam(nehareddyg)