CoDC: Unified Diffusion and Classification for Enhanced Class-Incremental Learning
In class-incremental learning (CIL), a model must learn new classes while retaining previous knowledge without storing all historical data. Generative replay mitigates catastrophic forgetting by synthesizing old class samples, but conventional pipelines usually train separate generative and classifi...
Uloženo v:
| Hlavní autoři: | , , , |
|---|---|
| Médium: | Artigo |
| Jazyk: | Inglês |
| Vydáno: |
MDPI AG
2026-06-01
|
| Edice: | Applied Sciences |
| Témata: | |
| On-line přístup: | https://www.mdpi.com/2076-3417/16/12/6035 |
| Tagy: |
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
