Memory Matching Networks for One-Shot Image Recognition

Cai, Qi; Pan, Yingwei; Yao, Ting; Yan, Chenggang; Mei, Tao

doi:10.1109/cvpr.2018.00429

Cited by 243 publications

(119 citation statements)

References 18 publications

(20 reference statements)

Supporting

Mentioning

119

Contrasting

Order By: Relevance

“…Briefly, optimization-based methods usually associated with the concept of meta-learning/learning to learn [7,24], e.g., learning a meta-optimizer [18] or taking some wise optimization strategies [6,28,29], to better and faster update the model for new tasks. Memory-based methods generally introduced memory components to accumulate experience when learning old tasks and generalize them when performing new tasks [3,16,19]. Our experimental results show that our method outperforms them without the need for updating the model for new tasks or introducing complicated memory structure.…”

Section: Related Workmentioning

confidence: 92%

“…2 Trained with 30-way 15 queries per episode task. 3 Our reimplementation of RN [23]. are optimization-based, and the fourth method (MMNets) is memory-based.…”

Section: -Shot 5-shotmentioning

confidence: 99%

“…The result of our reimplementation of RN [23] is better than the reported because our 2 × 2 max-pooling layers are applied in the last two layers but not the first two, and avoid premature loss of information. Compared with the optimization-based [6,18,28] and memory-based methods [3], our proposed PARN achieves better accuracies without the need for updating the model for new tasks or introducing complicated memory structure. As for metric-based methods, after combining DFE and DCA, PARN improves RN from 51.64% to 55.22% in the 1-shot task and 66.08% to 71.55% in the 5-shot task, and defeats all the other metric-based methods by a clear margin.…”

Section: -Shot 5-shotmentioning

confidence: 99%

See 2 more Smart Citations

PARN: Position-Aware Relation Networks for Few-Shot Learning

Guo

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

Few-shot learning presents a challenge that a classifier must quickly adapt to new classes that do not appear in the training set, given only a few labeled examples of each new class. This paper proposes a position-aware relation network (PARN) to learn a more flexible and robust metric ability for few-shot learning. Relation networks (RNs), a kind of architectures for relational reasoning, can acquire a deep metric ability for images by just being designed as a simple convolutional neural network (CNN) [23]. However, due to the inherent local connectivity of CNN, the CNN-based relation network (RN) can be sensitive to the spatial position relationship of semantic objects in two compared images. To address this problem, we introduce a deformable feature extractor (DFE) to extract more efficient features, and design a dual correlation attention mechanism (DCA) to deal with its inherent local connectivity. Successfully, our proposed approach extents the potential of RN to be position-aware of semantic objects by introducing only a small number of parameters. We evaluate our approach on two major benchmark datasets, i.e., Omniglot and Mini-Imagenet, and on both of the datasets our approach achieves state-of-theart performance with the setting of using a shallow feature extraction network. It's worth noting that our 5-way 1-shot result on Omniglot even outperforms the previous 5-way 5shot results.

show abstract

Section: Related Workmentioning

confidence: 92%

“…2 Trained with 30-way 15 queries per episode task. 3 Our reimplementation of RN [23]. are optimization-based, and the fourth method (MMNets) is memory-based.…”

Section: -Shot 5-shotmentioning

confidence: 99%

Section: -Shot 5-shotmentioning

confidence: 99%

See 1 more Smart Citation

PARN: Position-Aware Relation Networks for Few-Shot Learning

Guo

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

show abstract

“…A prototypical network [22] learns a prototype for each category, so that the examples discriminatively cluster around the prototypes corresponding to each category. Some studies [20], [25] developed memory modules that store useful information from training examples and exploited their memories when testing them. On the contrary, the relationship between activations and weights was established by [21].…”

Section: Related Workmentioning

confidence: 99%

Few-Shot Learning With Geometric Constraints

Jung

Lee

2020

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

In this paper, we consider the problem of few-shot learning for classification. We assume a network trained for base categories with a large number of training examples, and we aim to add novel categories to it that have only a few, e.g., one or five, training examples. This is a challenging scenario because (1) high performance is required in both the base and novel categories, and (2) training the network for the new categories with few training examples can contaminate the feature space trained well for the base categories. To address these challenges, we propose two geometric constraints to fine-tune the network with a few training examples. The first constraint enables features of the novel categories to cluster near the category weights, and the second maintains the weights of the novel categories far from the weights of the base categories. By applying the proposed constraints, we extract discriminative features for the novel categories while preserving the feature space learned for the base categories. Using public datasets for few-shot learning that are subsets of ImageNet, we demonstrate that the proposed method outperforms prevalent methods by a large margin.

show abstract

“…Furthermore, unlike MAML, we do not limit the length of the item-consumption history. We extend the idea of the matching network, which is one of the most famous meta-learning algorithms and shows good performance even when the length of the support set (i.e., length of the item-consumption history) is not fixed [2].…”

Section: Meta-learned User Preference Estimatormentioning

confidence: 99%

MeLU

Lee

Jang

et al. 2019

Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

212

View full text Add to dashboard Cite

This paper proposes a recommender system to alleviate the coldstart problem that can estimate user preferences based on only a small number of items. To identify a user's preference in the cold state, existing recommender systems, such as Netflix, initially provide items to a user; we call those items evidence candidates.Recommendations are then made based on the items selected by the user. Previous recommendation studies have two limitations:(1) the users who consumed a few items have poor recommendations and (2) inadequate evidence candidates are used to identify user preferences. We propose a meta-learning-based recommender system called MeLU to overcome these two limitations. From metalearning, which can rapidly adopt new task with a few examples, MeLU can estimate new user's preferences with a few consumed items. In addition, we provide an evidence candidate selection strategy that determines distinguishing items for customized preference estimation. We validate MeLU with two benchmark datasets, and the proposed model reduces at least 5.92% mean absolute error than two comparative models on the datasets. We also conduct a user study experiment to verify the evidence selection strategy.

show abstract

Memory Matching Networks for One-Shot Image Recognition

Cited by 243 publications

References 18 publications

PARN: Position-Aware Relation Networks for Few-Shot Learning

PARN: Position-Aware Relation Networks for Few-Shot Learning

Few-Shot Learning With Geometric Constraints

MeLU

Contact Info

Product

Resources

About