Details

Autor(en) / Beteiligte

• Ahmed, Mohiuddin
• Islam, Sheikh Rabiul
• Anwar, Adnan
• Moustafa, Nour
• Pathan, Al-Sakib Khan

Titel

The Past, Present, and Prospective Future of XAI: A Comprehensive Review

Ist Teil von

• Explainable Artificial Intelligence for Cyber Security, 2022, Vol.1025, p.1-29

Ort / Verlag

Switzerland: Springer International Publishing AG

Erscheinungsjahr

2022

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• With the increasing growth and availability of data, Artificial Intelligence (AI) based black-box models have shown significant effectiveness to solve real-world and mission-critical problems in a wide range of domains such as finance, healthcare, security, natural language processing, transportation, military, and retail. However, the importance of interpretable and explainable decision-making processes within AI-based systems is becoming crucial to provide transparency and confidence among end-users from diverse backgrounds. Acceptance of these black-box models considerably depends on to which extent the users or technical personnel understand and trust the underlying mechanism. However, ML methods and models are getting more sophisticated and transparent progressively. Though the domain experts understand the mathematical principle, they still face difficulties expressing the mechanism for a wide audience. To address this issue, a recent research domain—Explainable AI (XAI), has emerged intending to generate the contextual explanatory model for practical deployment. In this chapter, we systematically review and examine the existing literature and contribution of XAI in different domains. The presented taxonomy contains our investigation on the necessity of explaining and contextualizing methods and models in the AI ecosystem. Moreover, our study includes prospects in this field and the possible consequences of more explainable and interpretable ML models concerning different domains. This critical taxonomy provides the opportunities and challenges in the area of XAI that serves as a reference to future AI researchers.