Insight into Connection

Inspiration

Our story begins with a passion for solving challenging problems that matter. By creating a system for analyzing graph data, we are able to provide solutions in the various realms of industry and tackle a wide range of problems.

What it does

Insight into Connection system builds the machine learning framework for predicting future edges in the graph based on the node features.

How we built it

We applied various graph analyses to get more insight into the nature of the graph and find the most discriminatory features that we used for prediction in a supervised learning setup.

Since a list of features associated with the nodes is categorical we applied one-hot-encoding and concatenate them to other features associated with their connection in the graph.

We applied the space reduction and applied graph convolution to get insights from each node about its neighbors. After

Challenges we ran into

We tried different approaches like Graph Attention Networks and Graph Sage, but these approaches required more time to implement than we had.

Accomplishments that we're proud of

We are especially proud of the teamwork and ideas that we managed to implement.

What we learned

We come to the conclusion that the analysis of the features combined with standard deep network works often better than more advanced models

What's next for Insight into Connection

We plan to expand on our knowledge and to be ready to solve the problems of the future

Built With

• pandasdataframe
• python
• tensorflow