Few-Shot Learning With Global Class Representations

By Tiange Luo, Aoxue Li, Tao A. Xiang, Weiran Huang, Liwei Wang. 2019

121 citations recorded by the source corpus.

In this paper, we propose to tackle the challenging few-shot learning (FSL) problem by learning global class representations using both base and novel class training samples. In each training episode, an episodic class mean computed from a support set is registered with the global representation via a registration module. This produces a registered global class representation for computing the classification loss using a query set. Though following a similar episodic training pipeline as existing meta learning based approaches, our method differs significantly in that novel class training samples are involved in the training from the beginning. To compensate for the lack of novel class training samples, an effective sample synthesis strategy is developed to avoid overfitting. Importantly, by joint base-novel class training, our approach can be easily extended to a more practical yet challenging FSL setting, i.e., generalized FSL, where the label space of test data is extended to both base and novel classes. Extensive experiments show that our approach is effective for both of the two FSL settings.

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

In Computer Vision and Pattern Recognition

In Meta & Continual Learning

Citations in the corpus

Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need? (2020, neutral): “49 ± n/a GCR [27] 64-64-64-64 53.”
Generalized and Incremental Few-Shot Learning by Explicit Learning and Calibration without Forgetting (2021, background): “For the state-ofthe-art comparison, we average over 600 episodes [40, 28, 9], for all other experiments over 100 episodes.”
Incremental Few-Shot Object Detection (2020, neutral): “Note that there are many more recent FSL methods [6, 26, 7, 17, 47, 55, 50, 27, 39, 11, 16] that produce better performance on image classification tasks than MAML.”
A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunities (2022, methodology): “Method Venue Type Experimental Dataset Description Query set External labeled set External unlabeled set Congealing [57] CVPR’00 Data Expansion MNIST DAD [62] ECCV’14 Data Expansion Sign Language* FTT [60] CVPR’16 Data Expansion TADB Low-shot-shrink-hallucinate [58] ICCV’17 Data Expansion ImageNet F”
Learning What Not to Segment: A New Perspective on Few-Shot Segmentation (2022, methodology): “On this basis, FSL approaches can be further subdivided into three branches: (i) metric-based [24, 49, 50], (ii) optimization-based [11, 19, 42], and (iii) augmentation-based [5, 6].”
Mutual CRF-GNN for Few-shot Learning (2021, methodology): “To validate the effectiveness of our MCGN compared with other methods [42, 56, 30] in few-shot learning, we fix the feature extractor and train MCGN on top of the feature extractor.”
Boosting Few-Shot Learning With Adaptive Margin Loss (2020, background): “vel classes with only a few labeled samples, FSL aims to learn a classiﬁer for the novel classes by learning a generic knowledge from the base classes. Recently, metric-based meta-learning approaches [8,14, 16,27,29] have shown the superior performance in solving the FSL problem, with attractive sim”
Learning to Segment the Tail (2020, background): “-balance the training samples across classes [16,3,11,6]; while the latter focuses on assigning different weights to adjust the loss function [18,40,47,7]. Some works on generalized few-shot learning [46,21] also deal with an extremely imbalanced dataset, extending the test label space of few-shot l”
Relational Generalized Few-Shot Learning (2019, background): “propagating labels from class-level to instance-level graphs [51], and learning generative models to synthesize additional image features for the few-shot classes [49] or additional samples [22].”
Few-Shot Learning With a Strong Teacher (2021, unknown): “For example, based on the meta-trained feature extractor, the few-shot learner classiﬁes the query instances with (adapted) embeddings [3], [10], [13], [28], [55], [56], [57].”
Bowtie Networks: Generative Modeling for Joint Few-Shot Recognition and Novel-View Synthesis (2020, background): “, 2017), including the recent efforts on leveraging generative models (Li et al., 2015; Wang et al., 2018; Schwartz et al., 2018; Zhang et al., 2018; Tsutsui et al., 2019; Chen et al., 2019b; Li et al., 2019; Zhang et al., 2019; Sun et al., 2019).”