PlaStIL: Plastic and Stable Exemplar-Free Class-Incremental Learning - CEA - Commissariat à l’énergie atomique et aux énergies alternatives Accéder directement au contenu
Poster De Conférence Année : 2023

PlaStIL: Plastic and Stable Exemplar-Free Class-Incremental Learning

Résumé

Plasticity and stability are needed in class-incremental learning in order to learn from new data while preserving past knowledge. Due to catastrophic forgetting, finding a compromise between these two properties is particularly challenging when no memory buffer is available. Mainstream methods need to store two deep models since they integrate new classes using fine-tuning with knowledge distillation from the previous incremental state. We propose a method which has similar number of parameters but distributes them differently in order to find a better balance between plasticity and stability. Following an approach already deployed by transfer-based incremental methods, we freeze the feature extractor after the initial state. Classes in the oldest incremental states are trained with this frozen extractor to ensure stability. Recent classes are predicted using partially fine-tuned models in order to introduce plasticity. Our proposed plasticity layer can be incorporated to any transfer-based method designed for exemplar-free incremental learning, and we apply it to two such methods. Evaluation is done with three large-scale datasets. Results show that performance gains are obtained in all tested configurations compared to existing methods.
Fichier principal
Vignette du fichier
PlaStIL_CoLLAs2023-1.pdf (721.19 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

cea-04488070 , version 1 (04-03-2024)

Identifiants

Citer

Gregoire Petit, Adrian Popescu, Eden Belouadah, David Picard, Bertrand Delezoide. PlaStIL: Plastic and Stable Exemplar-Free Class-Incremental Learning. Second Conference on Lifelong Learning Agents - CoLLAs 2023, Aug 2023, Montréal, Canada. Proceedings of Machine Learning Research (PMLR), Second Conference on Lifelong Learning Agents - CoLLAs 2023, 232, pp.399-414, 2023, ⟨10.48550/arXiv.2209.06606⟩. ⟨cea-04488070⟩
8 Consultations
7 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More