Fuzzy Cognitive Maps and Explainable Artificial Intelligence: a critical perspective
Session
Computer Science and Communication Engineering
Description
There is a lot of discussion regarding the interpretability and explain- ability of modern artificial intelligence methodologies, especially in applications
such as medical imaging. Scientists argue that the most vital drawback of com- plex algorithms is their behaviour as black boxes. It is agreed that applying the
newly developed methods in industry, medicine, agriculture, and other modern fields, such as the Internet of Things, requires the trustfulness of the systems from the users. Users are always entitled to know why and how each method made a decision and which factors played a role. Otherwise, they will always be wary of using new techniques. Fuzzy Cognitive Maps are an evolving computational method to model human knowledge, provide decisions handling uncertainty, and are the core of many modern intelligent systems. Numerous studies in various fields employ FCMs, which report top performance, sometimes proving to be superior to several Machine Learning models. In this work, we analyze the nature of FCMs in terms of their trust, transferability, causality, informativeness, and transparency, providing the reader with several success stories that reveal the suitability of FMCs in many domains.
Keywords:
Fuzzy Cognitive Maps; Explainability; Explainable Artificial Intel- ligence; Interpretability
Proceedings Editor
Edmond Hajrizi
ISBN
978-9951-550-50-5
Location
UBT Kampus, Lipjan
Start Date
29-10-2022 12:00 AM
End Date
30-10-2022 12:00 AM
DOI
10.33107/ubt-ic.2022.281
Recommended Citation
Athanasoula, Ifigeneia; Apostolopoulos, Ioannis D.; and Groumpos, Peter P., "Fuzzy Cognitive Maps and Explainable Artificial Intelligence: a critical perspective" (2022). UBT International Conference. 290.
https://knowledgecenter.ubt-uni.net/conference/2022/all-events/290
Fuzzy Cognitive Maps and Explainable Artificial Intelligence: a critical perspective
UBT Kampus, Lipjan
There is a lot of discussion regarding the interpretability and explain- ability of modern artificial intelligence methodologies, especially in applications
such as medical imaging. Scientists argue that the most vital drawback of com- plex algorithms is their behaviour as black boxes. It is agreed that applying the
newly developed methods in industry, medicine, agriculture, and other modern fields, such as the Internet of Things, requires the trustfulness of the systems from the users. Users are always entitled to know why and how each method made a decision and which factors played a role. Otherwise, they will always be wary of using new techniques. Fuzzy Cognitive Maps are an evolving computational method to model human knowledge, provide decisions handling uncertainty, and are the core of many modern intelligent systems. Numerous studies in various fields employ FCMs, which report top performance, sometimes proving to be superior to several Machine Learning models. In this work, we analyze the nature of FCMs in terms of their trust, transferability, causality, informativeness, and transparency, providing the reader with several success stories that reveal the suitability of FMCs in many domains.