The Digital Democracy project (www.digitaldemocracy.org) is providing a dataset of screen captures from the videos of legislative hearings that took place in the state of California (and possibly in some other states) since 2015.  The images contain faces of different state legislators and also include some other labelling information. 

Your goal is to

- Build a machine learning pipeline that analyzes and properly classifies the images from the Digital Democracy dataset. 
- Use Convolutional Neural Networks and Google’s TensorFlow library as part of your machine learning pipeline.
- Measure the accuracy of your classification methodology. A Kaggle competition will be set up for this data set.  Attempt to beat the baseline accuracy of about 89%.
- Write a comprehensive report detailing your work. For this assignment your report shall read as a paper you are intending to submit and publish at a conference or a workshop.