Complementing Representation Deficiency in Few-shot Image Classification: A Meta-Learning Approach

21 Jul 2020  ·  Xian Zhong, Cheng Gu, Wenxin Huang, Lin Li, Shuqin Chen, Chia-Wen Lin ·

Few-shot learning is a challenging problem that has attracted more and more attention recently since abundant training samples are difficult to obtain in practical applications. Meta-learning has been proposed to address this issue, which focuses on quickly adapting a predictor as a base-learner to new tasks, given limited labeled samples. However, a critical challenge for meta-learning is the representation deficiency since it is hard to discover common information from a small number of training samples or even one, as is the representation of key features from such little information. As a result, a meta-learner cannot be trained well in a high-dimensional parameter space to generalize to new tasks. Existing methods mostly resort to extracting less expressive features so as to avoid the representation deficiency. Aiming at learning better representations, we propose a meta-learning approach with complemented representations network (MCRNet) for few-shot image classification. In particular, we embed a latent space, where latent codes are reconstructed with extra representation information to complement the representation deficiency. Furthermore, the latent space is established with variational inference, collaborating well with different base-learners, and can be extended to other models. Finally, our end-to-end framework achieves the state-of-the-art performance in image classification on three standard few-shot learning datasets.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Few-Shot Image Classification CIFAR-FS 5-way (1-shot) MCRNet-RR Accuracy 73.8 # 29
Few-Shot Image Classification CIFAR-FS 5-way (1-shot) MCRNet-SVM Accuracy 74.7 # 26
Few-Shot Image Classification CIFAR-FS 5-way (5-shot) MCRNet-RR Accuracy 85.2 # 29
Few-Shot Image Classification CIFAR-FS 5-way (5-shot) MCRNet-SVM Accuracy 86.8 # 24
Few-Shot Image Classification FC100 5-way (1-shot) MCRNet-SVM Accuracy 41 # 19
Few-Shot Image Classification FC100 5-way (1-shot) MCRNet-RR Accuracy 40.7 # 20
Few-Shot Image Classification FC100 5-way (5-shot) MCRNet-SVM Accuracy 57.8 # 17
Few-Shot Image Classification FC100 5-way (5-shot) MCRNet-RR Accuracy 56.6 # 20
Few-Shot Image Classification Mini-Imagenet 5-way (1-shot) MCRNet-SVM Accuracy 62.53 # 61
Few-Shot Image Classification Mini-Imagenet 5-way (1-shot) MCRNet-RR Accuracy 61.32 # 66
Few-Shot Image Classification Mini-Imagenet 5-way (5-shot) MCRNet-SVM Accuracy 80.34 # 46
Few-Shot Image Classification Mini-Imagenet 5-way (5-shot) MCRNet-RR Accuracy 78.16 # 54

Methods


No methods listed for this paper. Add relevant methods here