In this paper, a basic idea of Meta-learning and its applications are given. Meta-learning is a derivative of machine learning. It is commonly referred to as “learning to learn”. Simply this meta-learning means the learning model that gets knowledge through another learning model or algorithm. The area of meta-learning is growing rapidly year by year because of its ability to get precise results for a given input data in machine learning. There are several machine learning algorithms at present, the role of meta-learning is to understand which algorithm is most suitable for a specific dataset. Converse to the traditional way approach to artificial intelligence where tasks are resolved from the start with the help of established learning models, the main objective of meta-learning is to enhance the model automatically by giving instructions from the knowledge which is gained by observing various other learning algorithms. The meta-learning algorithm alters the existing algorithms for a particular problem and tries to get the best result out of it, this algorithm takes the outcome of the other algorithms as input. It is basically used to refine and upgrade the performance of the learning algorithms by altering some features of the algorithm. This algorithm gives accurate results as it takes the input data from the out data of the other algorithms.