The technological growth in our day to day life and its integration advancements in the communication network to activate a seamless communication have led to digital transformation in almost all applications. This causes a huge set of digital information conveyance via email, audio and video calls connecting people at all times. These data that are presently conveyed with the aid of the optic fiber communication technology would become outdated in the future years due to the growing demands of the digital information due to its intrinsic nonlinear effects. The machine learning appears as the promising technology to handle the complexities to be faced in the future systems by identify novel methodologies and utilizing available resources. The paper is to present the review on the nonlinearities experienced in the optical fiber and the promising solution provided by the machine learning techniques to enhance the capabilities of the optical fiber communication.