Optical Fiber Radio (RoF) systems based on OFDM meet the needs of high transmission and reception speeds, as well as offering greater reliability in the system. These systems are exposed to various disturbances, such as the thermal and shot noise of the photodetector, the amplified emission of optical links, and the relative phase intensity in the optical oscillator. To partially address these drawbacks, techniques such as multi-carrier modulation (OFDM), pilot-assisted equalization (PAE), and typical filters have been used. Recently, Extreme Learning Machines (ELM) have been employed instead of classic digital signal processing in RoF-OFDM systems to tackle physical limitations. ELMs are learning algorithms that have low latency rates and the ability to process large volumes of data. This article presents a review and comparison of the main research studies that have utilized ELM. It should be noted that ELM-C achieved the shortest equalization time in most cases compared to other algorithms.