Graph Structures and Machine Learning

• Challenges of Machine Learning: It is not trivial to handle data structures other than vectors in Machine Learning (does it mean non-mainstream?).
  • For example, graphs.
  • If we can handle graphs, we can use them for various things such as friend relationships or compounds (just an example).
  • Graph kernels.
• Graph Deep Learning.
  • Using Neural Networks for feature extraction from graph structures.
• Graph Convolution.
  • Similar to CNN, it involves incorporating information from neighboring vertices and compressing it. (Expert in Information Science)