Notes

training set and validation set

both have labels.

training set is used for model creation

validation set is to test the model

Training MNIST without transfer learning

Testing set

this is the data used for inference

Story behind Resnet

Make your own resnet

Visualizing the Loss function of neural network

read this research paper

Dense net are very memory intensive. they have very few parameters

They are an extension of Resnets. They tend to work really well on small datasets

GAN

Generative adversarial network

generator and critic