Answers are grounded in this paper and cite their sources.
Machine learning is the key technology of artificial intelligence, which uses learning and training data model to find the problem to achieve the law. In practical applications, there is always a difference between the input data of the model and the training data. Based on the biased training data, overfitting will occur, which will cause the machine learning to fail to achieve the expected goals. Generalization is the process of ensuring that the model can fully reflect the characteristics of the actual input data. Therefore, the ability of the machine learning model depends to a large extent on the effectiveness of generalization. In order to fully understand the learning and generalization model, we conducted a study by the polynomial curve fitting. First, based on the training data, we analyzed using matrix theory and analytical solutions derived polynomial fit. Then, numerical and maximum likelihood theoretical analysis was carried out for the overfitting problem. Finally, combined with the objective function, a typical regularization method to overcome overfitting and improve generalization ability is elaborated.
No takes yet. Share an insight, caveat, or question.
Bu et al. (2020) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: