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