Nowadays Machine Learning technologies rely just on correlations between the different features. Although, this approach can possibly lead to wrong conclusions since correlation does not necessarily imply causation. As part of this research study, I created and deployed on Amazon Web Services (AWS) a suite of Agent-Based and Compartmental Models in order to simulate epidemic diseases developments in different types of communities. All the code used as part of this project is available on my Github account and some extras are additionally available at this link.
Abstract
Nowadays Machine Learning models, are able to learn from data by identifying patterns in large datasets. Although, humans might be able to perform a same task after just examining a few examples. This is possible thanks to the inherit humans ability to understand causal relationships and use inductive inference in order to assimilate new information about the world. Creating models able to demonstrate causal reasoning would therefore open a whole new world of opportunities in Artificial Intelligence research. Causality arises naturally in our daily life every time we ask ourselves any type of interventional or retrospective question (eg. What if I take this action? What if I acted differently?). Causality has been researched and used for many years in statistics but not in great depth in Artificial Intelligence. Identifying useful connections between these two different ambit could therefore play a vital role in making a breakthrough towards creating intelligent systems. Enabling Machine Learning models to be more easily examinable to gain insights of their decision making processes could in fact facilitate adoption of these kind of technologies in fields such as medicine, surveillance and recruitment.
Google Scholar
Publication Download
Video Demonstration
Contacts
If you want to keep updated with my latest articles and projects follow me on Medium and subscribe to my mailing list. These are some of my contacts details:
