Playing with Neural Networks

Here I have experimented with various established datasets and carried out my own implementation of networks and have reported the accuracies.

1. CIFAR 10 : Using my own implementation of VGG13, I was able to get an test accuracy of 88.55 % and a test loss of 0.4972. (PyTorch)( Source code can be found here)
2. Flower:  Using the flower dataset, I was able to  get test accuracy of 81.81 % and a test loss of 0.777. (PyTorch)(Source code can be found here)
3. Fashion MNIST: I was able to get a test accuracy of 91.64 % and loss of 0.27. (keras)(Source code can be found here)
4. MNIST: I was able to get a test accuracy of 99.06 % and loss of 0.0440. (keras)(Source code can be found here)