MicroRNAs (miRNAs) are non-coding genes that are known to control gene expression by binding with messenger RNAs (mRNAs), mostly at the post-transcriptional level. The ability to fine-tune the regulation of gene expression has placed miRNAs at the centre of cancer research, where they have been reported to affect processes like tumour growth, metastasis, apoptosis, and genome stability. Due to this, understanding how miRNAs behave in cancer has gained great importance for identifying potential biomarkers and developing more precise therapies. In this chapter, we take a closer look at how various computational methods, especially those involving machine learning (ML), are enhancing our ability to identify miRNAs and predict their target genes. It starts by describing the biology of the miRNAs and how variations in expression levels contribute to cancers. Traditionally used tools like TargetScan, miRanda, and RNAhybrid are covered, along with their limitations of accuracy and consistency. Further, we focus on ML, which offers a more versatile approach by integrating large biological data, including sequence patterns, structural features, and gene expression profiles. Different ML models, ranging from supervised to deep learning, have been discussed for their strengths and use cases in bioinformatics, genomics, and cancer research. Recent studies applying ML to real-world cancer datasets further demonstrate its translational potential in identifying clinically relevant miRNAs and their gene targets. Though promising, ML-based methods are prone to perennial problems, such as dealing with high-dimensional data, missing values, and, most importantly, overfitting, which is associated with small sample sizes. This chapter highlights the importance of robust feature choice, stable model assessment, and interdisciplinary collaboration towards creating models that are not only predictive but also biologically relevant.