Few-shot learning in the era of Big Data: a literature review on perspectives and challenges

The advancement of Big Data has brought a large volume of data that has enabled the use of machine learning techniques for decision-making in different fields. However, the effectiveness of these models depends on the availability of large amounts of data, raising the challenge of dealing with learning from few samples. Learning from few samples is important in many applications when large volumes are not available, such as opinion surveys, due to the challenges of data collection, primarily due to lack of engagement and participation in questionnaires. This approach can facilitate or minimize the use of opinion surveys. However, there are challenges that need to be overcome to achieve accurate performance in few-shot learning tasks, such as the selection of relevant samples, and the choice of appropriate training methods, among others. In light of this, we will discuss the perspectives and challenges of learning from few samples in the era of Big Data. A review of Few-shot learning techniques and their applications will be conducted as an alternative to deal with few samples. Reviewing these techniques and their application on limited datasets can provide valuable insights for improving machine learning models in different application domains.