A little about my project.
This summer, I will be working in Dr. Koyejo's lab at the University of Illinois at Urbana-Champaign. The project I will be working on stems from the idea of Meta Learning. Meta Learning is the idea of learning to learn, so we are basically trying to build a Deep Learning model that learns to learn. To do this we are going to build a data set of models fitted to the cifar10 dataset. With this dataset we are going to attempt to predict the generalization or test error that the model would give by using it as an input to an RNN. If we are successful in predicting the generalization error, then we will be able to tell how well a model will do on a test set without actually testing it. We may also be able to use this information to generate models that have a low generalization error.