Change Password
  • Content format error The content can not be blank
  • Content format error The content can not be blank
  • Content format error The content can not be blank
The Previous Article:

XAI:Graph Learning & Graph Embedding

By Ricky Sun and Victor Wang

In our previous essays, we talked about graph data structures and database query languages, we also showcased broad-spectrum real-world applications that can be best powered with graph technologies. Social networks dwell on graphs that best model how people follow and befriend each other; biotech and pharmaceutical companies leverage graphs to understand protein interactions and chemical compounds efficacies; supply chains, telco networks, power grids are naturally presented as graphs. Many industries are looking to graphs to help with their businesses.

In the big-data era, more and more of these aforementioned businesses are interested in machine learnings to boost their predictability, some have been using deep learning and specifically varied neural networks to extract more predictive powers. There are 3 major problems lingering around though:

  • Siloed Systems within AI Eco-system
  • Low-Performance AI
  • Black-Box AI

Graph-0: AI Learning Techniques and Their Performance & Explainability

 

Graph-0 illustrates the 3 major problems well. First of all, there are quite a few learning techniques, but they are NOT offered in a one-stop-shop fashion; most AI users and programmers are dealing with multiple siloed systems, software and hardware-wise, which means a lot of data ETL are needed all the time (we’ll give detailed examples to illustrate how serious the problem is next). Secondly, AI (deep/machine) learning seems to be facing the dilemma of performance and explainability contradictory to each other, you can NOT have the best of both, and needless to say that as we go deeper into the deep-learning and ANN/CNN space, things are so black-box that human decision-makers are NOT okay with this status-quo. Lastly but certainly not the least important problem is most of these learning processes are NOT high-performance, and the data preparation, ETL, training, and application against production data can be a very complex and lengthy process! All in all, these problems deserve elevated attention, and ideally, infrastructure-level innovation should be attained to address and solve these challenges.

Want to read more?
The Next Article: