Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning

By Kai Zhu, Yang Cao, Wei Zhai, Jie Cheng, Zhengjun Zha. 2021

188 citations recorded by the source corpus.

Few-shot class-incremental learning is to recognize the new classes given few samples and not forget the old classes. It is a challenging task since representation optimization and prototype reorganization can only be achieved under little supervision. To address this problem, we propose a novel incremental prototype learning scheme. Our scheme consists of a random episode selection strategy that adapts the feature representation to various generated incremental episodes to enhance the corresponding extensibility, and a self-promoted prototype refinement mechanism which strengthens the expression ability of the new classes by explicitly considering the dependencies among different classes. Particularly, a dynamic relation projection module is proposed to calculate the relation matrix in a shared embedding space and leverage it as the factor for bootstrapping the update of prototypes. Extensive experiments on three benchmark datasets demonstrate the above-par incremental performance, outperforming state-of-the-art methods by a margin of 13%, 17% and 11%, respectively.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Han-Jia Ye
2. De-chuan Zhan
3. Da-Wei Zhou
4. Jun Zhu
5. Hang Su
6. Xingxing Zhang
7. Liyuan Wang
8. Dacheng Tao

In Computer Vision and Pattern Recognition

1. Han-Jia Ye
2. De-chuan Zhan
3. Dacheng Tao
4. Jun Zhu
5. Da-Wei Zhou
6. Luc Van Gool
7. Hang Su
8. Bernt Schiele

In Meta & Continual Learning

1. Han-Jia Ye
2. De-chuan Zhan
3. Da-Wei Zhou
4. Liyuan Wang
5. Xingxing Zhang
6. Hang Su
7. Jun Zhu
8. Fu-Yun Wang

Citations in the corpus

• A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, methodology): “Under this framework, representative strategies can be conceptually separated into (1) obtaining compatible and extensible representations from the base classes, such as by enforcing the representations compatible with simulated incremental tasks (SPPR [511], LIMIT [505]), reserving the”
• S3C: Self-Supervised Stochastic Classifiers for Few-Shot Class-Incremental Learning (2023, background): “The works proposed in [9,8,1,10,48,21,32] focus on learning robust manifolds by regular-izing feature space representations.”
• Few-Shot Class-Incremental Learning from an Open-Set Perspective (2022, neutral): “Most papers about FSCIL [4,25,7,23,27,5] follow the task setting proposed by Tao et al.”
• Few-Shot Class-Incremental Learning via Entropy-Regularized Data-Free Replay (2022, background): “[32] propose a self-promoted prototype refinement mechanism to explicitly consider the dependencies among classes, which results in an extensible feature representation.”
• Few-Shot Class-Incremental Learning by Sampling Multi-Phase Tasks (2022, background): “[24] only consider extending the model in a single phase, while FSCIL is a dynamic process with multiple phases.”
• Forward Compatible Few-Shot Class-Incremental Learning (2022, methodology): “Besides, we also compare to current SOTA FSCIL algorithms: TOPIC [43], SPPR [69], Decoupled-DeepEMD/Cosine/NegCosine [26, 47, 59] and CEC [60].”
• Self-Sustaining Representation Expansion for Non-Exemplar Class-Incremental Learning (2022, background): “CIL [32, 45] is the most difficult scenery in continual learning and has received more attention recently.”
• Representation Compensation Networks for Continual Semantic Segmentation (2022, background): “The knowledge can be stored by multiple types, like examples [4, 6, 9, 11, 73, 83], prototypes [35, 106, 107], generative networks [60], etc.”
• Continual Few-shot Relation Learning via Embedding Space Regularization and Data Augmentation (2022, methodology): “Indeed, in relation to CFRL,087 Zhang et al. (2021), Zhu et al. (2021) and Chen and088 Lee (2021) recently introduce methods for incre-089 mental few-shot learning in Computer Vision.090 Based on the observation that the learning of091 emerging few-shot tasks may result in distorted fea-092 ture…”
• Co-Transport for Class-Incremental Learning (2021, background): “Class-incremental learning (CIL) is now a popular topic in the machine learning community [6, 9, 15, 25, 39, 47, 62, 68, 70].”
• Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class Incremental Learning (2023, methodology): “Current studies widely rely on evolving the prototypes (Zhang et al., 2021; Zhu et al., 2021a) or sophisticated designs of loss and regularizer (Ren et al., 2019; Hou et al., 2019; Tao et al., 2020a; Joseph et al., 2022; Lu et al., 2022; Hersche et al., 2022; Chen & Lee, 2021; Aky¨urek et al.,…”
• Warping the Space: Weight Space Rotation for Class-Incremental Few-Shot Learning (2023, background): “There have been many attempts to solve class-incremental few-shot learning where only a few labeled samples are available in each task (Cheraghian et al., 2021b; Zhu et al., 2021; Zhou et al., 2022b; Cheraghian et al., 2021a).”
• Few-Shot Incremental Learning for Label-to-Image Translation (2022, methodology): “To alleviate the dependency on the amount of data for new tasks, advanced methods are proposed to solve fewshot incremental learning problems [11, 21, 29, 61].”
• A Survey on Few-Shot Class-Incremental Learning (2023, methodology): “To address these issues, Zhu et al. (2021a) proposes an incremental prototype learning scheme consisting of random episode selection and dynamic relation projection.”
• MetaFSCIL: A Meta-Learning Approach for Few-Shot Class Incremental Learning (2022, background): “[4] Zhu, Kai and Cao, Yang and Zhai, Wei and Cheng, Jie and Zha, Zheng-Jun.”