Literature / Related Work

We learned a lot with the help of our assigned TF, Camilo, who provided us many resources and helped us know what libraries and tools already existed. With his help, much of our research focused on understanding Kera’s capabilities and how to use it. We used documentation on keras.io and Stack Overflow extensively, but those are not listed below for the sake of conciseness.

Online Resources

We looked at some resources to understand how to use Keras for logistic regression and how to create our own data generator. Here are the resources we looked at for logistic regression.

  1. https://medium.com/@the1ju/simple-logistic-regression-using-keras-249e0cc9a970
  2. https://medium.com/@luwei.io/logistic-regression-with-keras-d75d640d175e
  3. https://stanford.edu/~shervine/blog/keras-how-to-generate-data-on-the-fly

We looked at several sources to understand how CNNs worked, some recommended by our TF, Camilo, and others we found on our own. Some of the sources we reviewed were:

  1. https://medium.com/@udemeudofia01/basic-overview-of-convolutional-neural-network-cnn-4fcc7dbb4f17
  2. https://adeshpande3.github.io/A-Beginner’s-Guide-To-Understanding-Convolutional-Neural-Networks/
  3. https://medium.com/@RaghavPrabhu/understanding-of-convolutional-neural-network-cnn-deep-learning-99760835f148

Much of our research focused on implementing the base CNN, but Camilo introduced us to several pre-trained CNNs available on https://keras.io/applications/. There is active research in developing the best pre-trained models, but we of course only used the ones that were already publicly available and made limited changes.

Resources from Class

  1. We watched the advanced section on CNNs.
  2. We looked at and based code off of the 209 HW 9 solution.