Open Access
Table 1
The comparative advantages of different AI models in terms of training data requirements, computational costs, and generalization capabilities.
| Model type | Training Data Requirement | Computational Cost | Generalization Capability | Primary Application Scenarios |
|---|---|---|---|---|
| Transformer | High | High | Strong | Sequential/spectral data, long-range dependency modeling |
| CNN | Medium | Medium | Medium | Image/structural data, local feature extraction |
| VAE | Medium-High | Medium-High | Strong | Latent space modeling, generating diverse structures |
| GAN | High | High | Medium | High-fidelity image/structure generation |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.