[Retracted] Review on the Application of Metalearning in Artificial Intelligence

Table 2

Specific analysis of the application of metalearning in the field of artificial intelligence.

Fields	Reasons for the rise	Specific application scenarios	Advantages
Few-shot learning	Limitations of dataset size	Face recognition [40]; Classification [41, 42]; Target detection [43, 44]; Video synthesis [40]	Low dependence on sample size; Strong generalization
Robot learning	The backwardness of robot operation skills	Imitation learning [49]; Cross-domain learning [50]; Quickly adapting online [51]	Improve the efficiency of autonomous learning by robots
Unsupervised learning	Poor performance of unsupervised learning algorithms	Distribution of unsupervised problems [52]; Noise training [53]	Simplifying unsupervised learning to supervised learning; Ability to learn from labeled data
Intelligent medicine	Slow progress in the medical field	Medical image processing [54]; Drug discovery [56]; Cancer detection [57]; Medical vision question answering [58]; Skin lesion segmentation tasks [60]	Predicting the specific behavior of molecules; Ability to learn to weigh support samples