Few-Shot Class-Incremental Learning by Sampling Multi-Phase Tasks

By Da-Wei Zhou, Han-Jia Ye, De-chuan Zhan. 2022

199 citations recorded by the source corpus.

New classes arise frequently in our ever-changing world, e.g., emerging topics in social media and new types of products in e-commerce. A model should recognize new classes and meanwhile maintain discriminability over old classes. Under severe circumstances, only limited novel instances are available to incrementally update the model. The task of recognizing few-shot new classes without forgetting old classes is called few-shot class-incremental learning (FSCIL). In this work, we propose a new paradigm for FSCIL based on meta-learning by LearnIng Multi-phase Incremental Tasks (Limit), which synthesizes fake FSCIL tasks from the base dataset. The data format of fake tasks is consistent with the ‘real’ incremental tasks, and we can build a generalizable feature space for the unseen tasks through meta-learning. Besides, Limit also constructs a calibration module based on transformer, which calibrates the old class classifiers and new class prototypes into the same scale and fills in the semantic gap. The calibration module also adaptively contextualizes the instance-specific embedding with a set-to-set function. Limit efficiently adapts to new classes and meanwhile resists forgetting over old classes. Experiments on three benchmark datasets (CIFAR100, miniImageNet, and CUB200) and large-scale dataset, i.e., ImageNet ILSVRC2012 validate that Limit achieves state-of-the-art performance.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Epidemiology, 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. Ziwei Liu
5. Hang Su
6. Jun Zhu
7. Qiwen Wang
8. Liyuan Wang

In Computer Vision and Pattern Recognition

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

In Epidemiology

1. Keon-Hee Park
2. Kyungwoo Song
3. Gyeong-Moon Park
4. Liyuan Wang
5. Jun Zhu
6. Hang Su
7. Han-Jia Ye
8. Xingxing Zhang

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. Qiwen 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”
• FOSTER: Feature Boosting and Compression for Class-Incremental Learning (2022, background): “Therefore, it is necessary to enable the model to continuously learn new categories, namely class-incremental learning [34,44,49].”
• Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class Incremental Learning (2023, background): “A popular choice is to use meta learning (Yoon et al., 2020; Chi et al., 2022; Zhou et al., 2022b).”
• Pre-trained Vision and Language Transformers are Few-Shot Incremental Learners (2024, methodology): “Feature and feature space-based methods [2, 12, 23, 30, 41, 42] enable the model to adapt to new classes better and improve the generalization ability of feature extractors for new classes.”
• A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning (2022, background): “There are other methods that do not fall into these two groups [24, 29, 23, 46, 61, 66], and we refer the readers to [10, 36, 62] for a holistic review.”
• Deep Class-Incremental Learning: A Survey (2023, background): “Specifically, [104], [106], [210], [211] address the few-shot class-incremental learning problem, where the model is required to be adapted to incoming few-shot classes.”
• A Survey on Few-Shot Class-Incremental Learning (2023, background): “In subsequent research, Zhou et al. (2022b) proposed LIMIT, which creates fake FSCIL tasks from the base dataset and obtains generalizable features through meta-learning from di ff erent fake tasks to prepare the model for real FSCIL tasks.”
• Warping the Space: Weight Space Rotation for Class-Incremental Few-Shot Learning (2023, methodology): “Additionally, we further compare our method with more recent works (Zhou et al., 2022b; Chi et al., 2022; Hersche et al., 2022; Liu et al., 2022) on mini ImageNet dataset.”
• Class-Incremental Learning: A Survey (2023, background): “Hence, FOSTER adds an extra model compression process by knowledge distillation [178]: Eq.”