HydraCIL decouples class-incremental learning by freezing the feature extractor and using prototype-guided multi-head classifiers to cut training time while matching SOTA accuracy on standard benchmarks.
Efficient single-step framework for incremental class learning in neural networks,
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
-
HydraCIL: Decoupled Class-Incremental Learning through Prototype-Guided Multi-Head Classifiers
HydraCIL decouples class-incremental learning by freezing the feature extractor and using prototype-guided multi-head classifiers to cut training time while matching SOTA accuracy on standard benchmarks.