EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
607. Causal Machine Learning: A Methodological Shift in Explainable AI for Decision Making
Invited abstract in session WA-11: Scenarios and foresight practices: Behavioural issues I, stream Behavioural OR.
Wednesday, 8:30-10:00Room: 12 (building: 116)
Authors (first author is the speaker)
| 1. | Rosa Taghikhah
|
| University of Sydney |
Abstract
Explainable AI (XAI) aims to enhance the transparency and interpretability of AI systems, a crucial need in decision analytics where understanding AI decision-making processes can significantly influence trust and operational adoption. Traditional XAI models often focus on one-to-one mappings between inputs and outputs, providing localized explanations for individual predictions. This approach, while useful, falls short in capturing the complex, interconnected relationships inherent in many real-world data sets. It particularly struggles with dynamic environments where feedback loops and interactions among variables play critical roles in shaping outcomes. These limitations highlight the need for more sophisticated methods that can unravel the intricate web of relationships and their causal effects on predictions. Causal machine learning offers a promising avenue for advancing explainability by focusing on the identification and understanding of causal relationships rather than mere correlations. Unlike traditional XAI models, causal approaches seek to model the entire system of variables and their interactions, including feedback loops. This enables a more comprehensive view of how changes in one part of the system can ripple through and affect other parts, providing insights into the mechanisms driving predictions. This enables insights into the mechanisms driving predictions.
Keywords
- Decision Support Systems
- Analytics and Data Science
- Graphs and Networks
Status: accepted
Back to the list of papers