Compositional Few-Shot Recognition with Primitive Discovery and Enhancing

By Yixiong Zou, Shanghang Zhang, José M. F. Moura, Yaowei Wang, Yonghong Tian. 2020

33 citations recorded by the source corpus.

Few-shot learning (FSL) aims at recognizing novel classes given only few training samples, which still remains a great challenge for deep learning. However, humans can easily recognize novel classes with only few samples. A key component of such ability is the compositional recognition that human can perform, which has been well studied in cognitive science but is not well explored in FSL. Inspired by such capability of humans, to imitate humans' ability of learning visual primitives and composing primitives to recognize novel classes, we propose an approach to FSL to learn a feature representation composed of important primitives, which is jointly trained with two parts, i.e. primitive discovery and primitive enhancing. In primitive discovery, we focus on learning primitives related to object parts by self-supervision from the order of image splits, avoiding extra laborious annotations and alleviating the effect of semantic gaps. In primitive enhancing, inspired by current studies on the interpretability of deep networks, we provide our composition view for the FSL baseline model. To modify this model for effective composition, inspired by both mathematical deduction and biological studies (the Hebbian Learning rule and the Winner-Take-All mechanism), we propose a soft composition mechanism by enlarging the activation of important primitives while reducing that of others, so as to enhance the influence of important primitives and better utilize these primitives to compose novel classes. Extensive experiments on public benchmarks are conducted on both the few-shot image classification and video recognition tasks. Our method achieves the state-of-the-art performance on all these datasets and shows better interpretability.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 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. Rui Li
2. Yuhua Li
3. Yuhua Li
4. Yixiong Zou
5. Ruixuan Li
6. Haichen Zhou
7. Shanghang Zhang
8. Shanghang Zhang

In Computer Vision and Pattern Recognition

1. Rui Li
2. Yuhua Li
3. Yuhua Li
4. Yixiong Zou
5. Ruixuan Li
6. Haichen Zhou
7. Shanghang Zhang
8. Shanghang Zhang

In Radiology, Nuclear Medicine and Imaging

1. Shanghang Zhang
2. Yixiong Zou

Citations in the corpus

• Margin-Based Few-Shot Class-Incremental Learning with Class-Level Overfitting Mitigation (2022, background): “As analyzed in [47] and [49], patterns (channels) with higher weights in the classification layer are more important, and the most important ones dominate the model decisions.”
• Revisiting Mid-Level Patterns for Cross-Domain Few-Shot Recognition (2020, unknown): “Few-shot learning methods can be roughly grouped into embedding based method [11, 41, 44, 55–57], meta-learning based method [2, 9, 28, 54], and hallucination based method [15, 43].”
• Compositional Few-Shot Class-Incremental Learning (2024, methodology): “This is because CPDE build primitives from the aspect of channels, but we build primitives based on image patches.”