Incremental few-shot learning via vector quantization in deep embedded space

By Kuilin Chen, Chi-Guhn Lee. 2021

121 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning, 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 Representation & Self-Supervised Learning

In Meta & Continual Learning

Citations in the corpus

S3C: Self-Supervised Stochastic Classifiers for Few-Shot Class-Incremental Learning (2023, methodology): “The works proposed in [9,8,1,10,48,21,32] focus on learning robust manifolds by regular-izing feature space representations.”
Towards Label-Efficient Incremental Learning: A Survey (2023, methodology): “Prominent examples include SWAV [Caron et al. , 2020], MOCO [Chen et al. , 2020], and BarlowTwins [Zbontar et al. , 2021].”
Constrained Few-shot Class-incremental Learning (2022, background): “Various solutions have been proposed, such as exploiting a neural gas network [48], a graph attention network [56], semantic word embeddings [6], and vector quantization [5].”
Continual Few-shot Relation Learning via Embedding Space Regularization and Data Augmentation (2022, background): “However, astute readers can find that 633 even without data augmentation, ERDA outper- 634 forms IDLVQ-C significantly for all tasks (compare 635 ‘ERDA w.o. DA’ with the baselines in Table 1).”
Overcoming Catastrophic Forgetting in Incremental Few-Shot Learning by Finding Flat Minima (2021, methodology): “The study of incremental few-shot learning has just started [47, 41, 60, 9, 8, 34, 59].”
Subspace Regularizers for Few-Shot Class Incremental Learning (2021, methodology): “Variant: Memory Following past work (Chen & Lee, 2021) which performs incremental learning while retaining a small “memory” of previous samples M , we explore an alternative baseline approach in which we append append S ( t ) in Eq.”
A Survey of Quantization Methods for Efficient Neural Network Inference (2021, neutral): “In particular, the work of [1, 30, 74, 83, 116, 166, 176, 184, 248] clusters the weights into different groups and use the centroid of each group as quantized values during inference.”
Synthesized Feature based Few-Shot Class-Incremental Learning on a Mixture of Subspaces (2021, background): “[4] propose a nonparametric approach in deep embedded space.”
Warping the Space: Weight Space Rotation for Class-Incremental Few-Shot Learning (2023, methodology): “Several other works adopt different strategies to learn new classes well during incremental sessions without forgetting (Tao et al., 2020; Cheraghian et al., 2021a; Zhang et al., 2021; Kukleva et al., 2021; Chen & Lee, 2021; Aky¨urek et al., 2022).”
Few-Shot Incremental Learning for Label-to-Image Translation (2022, methodology): “Following the FSCIL setting, later works propose strategies such as vector quantization [6] and calibrated classifiers [21] to improve classification performance.”
Neural Collapse Inspired Feature-Classifier Alignment for Few-Shot Class Incremental Learning (2023, methodology): “However, the two goals rely on sophisticated loss functions or regularizers (Chen & Lee, 2021; Hersche et al., 2022; Aky ¨ urek et al., 2022), and are hard to attain simultaneously without qualiﬁcation.”
A Survey on Few-Shot Class-Incremental Learning (2023, methodology): “Chen and Lee (2021) uses a non-parametric method based on LVQ in deep embedding space.”
On the Soft-Subnetwork for Few-shot Class Incremental Learning (2022, methodology): “Furthermore, we compare other FSCIL methods such as iCaRL [37], Rebalance [19], TOPIC [43], IDLVQ-C [5], and F2M [42].”