A Two-Stage Approach to Few-Shot Learning for Image Recognition

By Debasmit Das, C. S. Lee. 2019

141 citations recorded by the source corpus.

This paper proposes a multi-layer neural network structure for few-shot image recognition of novel categories. The proposed multi-layer neural network architecture encodes transferable knowledge extracted from a large annotated dataset of base categories. This architecture is then applied to novel categories containing only a few samples. The transfer of knowledge is carried out at the feature-extraction and the classification levels distributed across the two training stages. In the first-training stage, we introduce the relative feature to capture the structure of the data as well as obtain a low-dimensional discriminative space. Secondly, we account for the variable variance of different categories by using a network to predict the variance of each class. Classification is then performed by computing the Mahalanobis distance to the mean-class representation in contrast to previous approaches that used the Euclidean distance. In the second-training stage, a category-agnostic mapping is learned from the mean-sample representation to its corresponding class-prototype representation. This is because the mean-sample representation may not accurately represent the novel category prototype. Finally, we evaluate the proposed network structure on four standard few-shot image recognition datasets, where our proposed few-shot learning system produces competitive performance compared to previous work. We also extensively studied and analyzed the contribution of each component of our proposed framework.

Subfields: Artificial Intelligence, Radiology, Nuclear Medicine and Imaging

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. Hexiang Hu
4. Tao Mei
5. Fei Sha
6. Nuno Vasconcelos
7. Da-Wei Zhou
8. Debasmit Das

In Radiology, Nuclear Medicine and Imaging

1. Han-Jia Ye
2. De-chuan Zhan
3. Hexiang Hu
4. Tao Mei
5. Gang Hua
6. Fei Sha
7. C. S. George Lee
8. Nuno Vasconcelos

Citations in the corpus

• FeTrIL: Feature Translation for Exemplar-Free Class-Incremental Learning (2022, neutral): “We present the main components of FeTrIL in the next subsections.”
• Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learning (2019, background): “in various domains such as visual recognition (Triantafillou et al. 2017; Lifchitz et al. 2019; Das and Lee 2020), domain adaptation (Dong and Xing 2018; Kang et al.”
• MetaSearch: Incremental Product Search via Deep Meta-Learning (2020, background): “More recently, they also proposed a two-stage framework for few-shot learning of image recognition [7] that achieved excellent performance on public datasets.”
• Few-Shot Open-Set Recognition Using Meta-Learning (2020, background): “Few-Shot Learning: Extensive research on few-shot learning [26, 4, 34, 30, 32, 27, 7, 23, 16, 2] has emerged in recent years.”
• Plant Disease Detection and Classification by Deep Learning—A Review (2021, background): “Das and Lee [96] proposed a two-stage multilayer neural network for the few-shot recognition of new categories and a detailed mathematical theory derivation process.”
• Leveraging the Feature Distribution in Transfer-based Few-Shot Learning (2020, background): “Recently, the trend evolved towards using well-thought-out transfer architectures (called backbones) [27, 7], typically trained on large available labelled datasets that differ in their content from the ones of the task.”
• Zero-shot Image Recognition Using Relational Matching, Adaptation and Calibration (2019, background): “data but also reduces the dimensionality of the feature space implicitly [17], thus alleviating the hubness problem.”