Matches in SemOpenAlex for { <https://semopenalex.org/work/W4381569594> ?p ?o ?g. }
Showing items 1 to 51 of
51
with 100 items per page.
- W4381569594 abstract "Few-shot class-incremental learning (FSCIL) aims to continually learn new classes using a few samples while not forgetting the old classes. The key of this task is effective knowledge transfer from the base session to the incremental sessions. Despite the advance of existing FSCIL methods, the proposed knowledge transfer learning schemes are sub-optimal due to the insufficient optimization for the model's plasticity. To address this issue, we propose a Random Episode Sampling and Augmentation (RESA) strategy that relies on diverse pseudo incremental tasks as agents to achieve the knowledge transfer. Concretely, RESA mimics the real incremental setting and constructs pseudo incremental tasks globally and locally, where the global pseudo incremental tasks are designed to coincide with the learning objective of FSCIL and the local pseudo incremental tasks are designed to improve the model's plasticity, respectively. Furthermore, to make convincing incremental predictions, we introduce a complementary model with a squared Euclidean-distance classifier as the auxiliary module, which couples with the widely used cosine classifier to form our whole architecture. By such a way, equipped with model decoupling strategy, we can maintain the model's stability while enhancing the model's plasticity. Extensive quantitative and qualitative experiments on three popular FSCIL benchmark datasets demonstrate that our proposed method, named Knowledge Transfer-driven Relation Complementation Network (KT-RCNet), outperforms almost all prior methods. More precisely, the average accuracy of our proposed KT-RCNet outperforms the second-best method by a margin of 5.26%, 3.49%, and 2.25% on miniImageNet, CIFAR100, and CUB200, respectively. Our code is available at https://github.com/YeZiLaiXi/KT-RCNet.git." @default.
- W4381569594 created "2023-06-22" @default.
- W4381569594 creator A5014825654 @default.
- W4381569594 creator A5037124719 @default.
- W4381569594 creator A5065592637 @default.
- W4381569594 creator A5079805574 @default.
- W4381569594 date "2023-06-19" @default.
- W4381569594 modified "2023-10-18" @default.
- W4381569594 title "Knowledge Transfer-Driven Few-Shot Class-Incremental Learning" @default.
- W4381569594 doi "https://doi.org/10.48550/arxiv.2306.10942" @default.
- W4381569594 hasPublicationYear "2023" @default.
- W4381569594 type Work @default.
- W4381569594 citedByCount "0" @default.
- W4381569594 crossrefType "posted-content" @default.
- W4381569594 hasAuthorship W4381569594A5014825654 @default.
- W4381569594 hasAuthorship W4381569594A5037124719 @default.
- W4381569594 hasAuthorship W4381569594A5065592637 @default.
- W4381569594 hasAuthorship W4381569594A5079805574 @default.
- W4381569594 hasBestOaLocation W43815695941 @default.
- W4381569594 hasConcept C119857082 @default.
- W4381569594 hasConcept C150899416 @default.
- W4381569594 hasConcept C154945302 @default.
- W4381569594 hasConcept C22367795 @default.
- W4381569594 hasConcept C2776214188 @default.
- W4381569594 hasConcept C2780735816 @default.
- W4381569594 hasConcept C41008148 @default.
- W4381569594 hasConcept C95623464 @default.
- W4381569594 hasConceptScore W4381569594C119857082 @default.
- W4381569594 hasConceptScore W4381569594C150899416 @default.
- W4381569594 hasConceptScore W4381569594C154945302 @default.
- W4381569594 hasConceptScore W4381569594C22367795 @default.
- W4381569594 hasConceptScore W4381569594C2776214188 @default.
- W4381569594 hasConceptScore W4381569594C2780735816 @default.
- W4381569594 hasConceptScore W4381569594C41008148 @default.
- W4381569594 hasConceptScore W4381569594C95623464 @default.
- W4381569594 hasLocation W43815695941 @default.
- W4381569594 hasOpenAccess W4381569594 @default.
- W4381569594 hasPrimaryLocation W43815695941 @default.
- W4381569594 hasRelatedWork W2623427976 @default.
- W4381569594 hasRelatedWork W2960456850 @default.
- W4381569594 hasRelatedWork W2961085424 @default.
- W4381569594 hasRelatedWork W4213299466 @default.
- W4381569594 hasRelatedWork W4281382123 @default.
- W4381569594 hasRelatedWork W4281645081 @default.
- W4381569594 hasRelatedWork W4294306704 @default.
- W4381569594 hasRelatedWork W4308262314 @default.
- W4381569594 hasRelatedWork W4318834068 @default.
- W4381569594 hasRelatedWork W4318957922 @default.
- W4381569594 isParatext "false" @default.
- W4381569594 isRetracted "false" @default.
- W4381569594 workType "article" @default.