Critical infrastructures, such as power and water networks, are vital for society and the economy. However, they are vulnerable to various disruptions such as component failures, cyber-attacks, and natural disasters. These disruptions can cascade across critical infrastructure networks (CINs), causing significant socioeconomic losses. Decision-makers face the challenge of protecting CINs before disruptions and restoring their functions afterward, considering interdependencies and uncertainties. Current methods struggle to model big data, complex interactions, and multilayer dependencies between CINs. Artificial intelligence (AI) and machine learning (ML) applications can be used to overcome these challenges, as they can model complex systems and discover data patterns representing a promising research trend that could benefit both private companies and governments. This article undertakes a comprehensive review of the literature on the applications of machine learning in improving the resilience of interdependent critical infrastructure systems (ICISs). The aim is to address the existing knowledge gap and dispersed research articles in this area, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol. The primary goal of this article is to assess the current state of ML applications in the ICISs resilience engineering field by examining the available literature, in order to discover future opportunities and trends. The findings are summarized, and potential future trends and opportunities are listed, aiming to inspire resilience engineering practitioners to explore these future directions in the field.