ABSTRACT
Highlights
Explaining XAI, taxonomy, conceptions, and aims in Deep learning.
Detailed principles Goals and Paths followed by XAI.
Explaining the XAI model and its intended level of explainability.
Black-box attacks on XAI, major challenges.
Suggested solutions for preventing Adversarial black-box attacks in cyber security.
Explainable machine learning refers to efforts to ensure that artificial intelligence programmers are clear about their goals and how they operate. Explainable artificial intelligence (XAI) is one of the hot research topics and the most expanding fields in deep learning. The main motive of the XAI approach is to provide a human-reasonable clarification for the deep learning model. Explainable AI has made substantial development over the last few years. In both academia and business, the goal of transforming these black-box models into transparent and interpretable algorithms has gained support. This work provides an overview of enduring research and recent contributions, but more research on XAI still needs to be accomplished. The review also gives an in-depth analysis of XAI by grouping all scientific research using a hierarchical structure that defines theories and conceptions relevant to explainability and XAI methodologies. This chapter also discusses some of the essential black-box cyber security attacks, their challenges, and various suggested solutions to overcome these adversarial attacks.
