Communication networks play a vital role in the technological infrastructure underpinning Internet traffic applications. Service providers and researchers worldwide are sparing no effort to increase the information capacity and security of telecommunication networks to support the demands of high-speed, reliable and secure emerging internet, data centre, cloud computing, 5G new radio and IoT systems, especially since the outbreak of Coronavirus. Applications such as intelligent transportation, signal processing ubiquitous low-latency connectivity and massive connected objects, have raised challenges for backbone and access networks that are often underpinned by optical, radio or hybrid networks. Artificial intelligent (AI) technologies appear an innovative and promising solution to cope with emerging challenges in optical/wireless/hybrid networks, in which the underlying physics, mathematics and optimisation of problems are non-deterministic to analyse or impossible to describe explicitly. In this proposed research, supervised, unsupervised and reinforcement learning techniques such as neural networks, clustering and regression will be exploited in optical/wireless/hybrid networks to mitigate stochastic distortions, to predict network conditions and to maximise network capacity. This DIOR proposal aims to unite optical/radio network research and AI technologies for tackling emerging challenges. This project aims to carry out world-leading research on building a machine learning-based communication platform to accelerate secure, intelligent and high-capacity communication networks.