Prototype Latent World Model Replay stores class prototypes as latent distributions and replays sampled states to achieve class-incremental learning without raw exemplars, raising LastAcc on Split CIFAR-100 from 4.55% to 31.64% (Inc5), 9.06% to 37.06% (Inc10), and 16.96% to 43.10% (Inc20).
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Prototype Latent World Model Replay for Class-Incremental Learning
Prototype Latent World Model Replay stores class prototypes as latent distributions and replays sampled states to achieve class-incremental learning without raw exemplars, raising LastAcc on Split CIFAR-100 from 4.55% to 31.64% (Inc5), 9.06% to 37.06% (Inc10), and 16.96% to 43.10% (Inc20).