Distributed large-scale Cyber-Physical Systems (CPSs) are complex, interconnected systems that integrate physical processes with computational elements. Such advanced systems are managed and controlled through innovative tools and techniques enabled by communication networks. However, CPSs still suffer from vulnerabilities to cyber threats, abnormal behaviors, system faults, and malicious activities, all of which can compromise their functionality and stability. Ensuring the security and resilience of such complex systems hinges on the (near) real-time detection and mitigation of cyber threats. Despite significant advancements, many existing anomaly detection and mitigation algorithms struggle to balance effectiveness and real-time performance requirements. This research aims to develop learning-enabled techniques that allow systems to learn from experience and adapt, thereby enhancing their ability to detect and respond to emerging cyber threats. Specifically, this work focuses on identifying threats, detecting compromised components, and designing control strategies to optimally reconfigure available, uncompromised resources in distributed CPSs to bolster resilience against malicious activities. Ultimately, the goal is to ensure continued CPS functionality in the face of cyber-attacks, minimizing the impact on operations. The outcomes of this research will contribute to K-12, undergraduate, and graduate levels of education by developing courses, workshops, and hands-on projects. This project aims to develop a general framework to enhance the security and resilience of distributed CPSs. The framework will focus on three key objectives: (i) Detection of Cyber-Attacks and Malicious Activities: to effectively identify cyber-attacks and malicious activities in a distributed CPS using novel algorithms based on reduced sparse time series transformer networks; (ii) Impact Analysis and Diagnostics: to analyze and assess the impacts of the existing threats using a new diagnostic methodology enabling the system to determine the types, locations, and other relevant characteristics of the attacks from cyber-attacks signatures; and (iii) Attack Mitigation and System Resilience: to utilize the obtained attacks information and design learning-enabled mitigation strategies that countermeasure cyber-attacks impacts in the system and maintain system functionality close to the normal. The proposed mitigation algorithm incorporates novel techniques to go beyond traditional resiliency and develop 'Anti-Fragility', a property that allows the system to not only recover from cyber-attacks but to improve its robustness and adaptability in response to similar future threats by leveraging prior attack experiences. To achieve this, the project will develop a resilient control strategy that employs fast learning deep reinforcement learning algorithms, augmented by automatic transfer learning techniques to reduce training and convergence time. These innovations will enable the system to adapt quickly and efficiently to new attacks, thereby maintaining operational functionality and continuously improving its resilience. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.