Analysis of a Class Incremental Learning Approximate Replay Method Based on Pre-trained Model
This paper expounds that by approximating the sampling of pre-trained model representations,which rapidly corrects the parameter deviations caused by catastrophic forgetting during the previous round of incremental learning.Experimental validation and evaluation demonstrate that,compared to traditional methods,this approach offers significant advantages in both performance and training speed.
NETL
NSTL
万方数据
