Course Project for CS249 - UCLA
In this work, which serves as the course project for the CS249 class, we explored different implementations and configurations for the conditional random fields and explored their effectiveness in a variety of ways to understand graphs and utilize the connections in making final judgments regarding node labels.
CORA dataset has been used in this work, with one of the notebooks dedicated to statistically analyze, preprocess, and prune the dataset.
Even compared to the state of the art methodologies such as graph convolutional networks, this approach yields acceptable results and demonstrates that in spite of all the advancements in the deep learning area, CRFs should not be ignored when it comes to designing architectures.