Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domain
This work addresses robustness issues in CNNs for computer vision applications, offering a novel data augmentation approach that improves generalization, though it is incremental as it builds on existing frequency domain insights.
The paper tackles the problem of convolutional neural networks (CNNs) being overly reliant on amplitude spectra, which are sensitive to noise and corruptions, by proposing amplitude-phase recombination data augmentation to enhance robustness. The method achieves state-of-the-art performance on generalization tasks like common corruptions, out-of-distribution detection, and adversarial attacks.
Recently, the generalization behavior of Convolutional Neural Networks (CNN) is gradually transparent through explanation techniques with the frequency components decomposition. However, the importance of the phase spectrum of the image for a robust vision system is still ignored. In this paper, we notice that the CNN tends to converge at the local optimum which is closely related to the high-frequency components of the training images, while the amplitude spectrum is easily disturbed such as noises or common corruptions. In contrast, more empirical studies found that humans rely on more phase components to achieve robust recognition. This observation leads to more explanations of the CNN's generalization behaviors in both robustness to common perturbations and out-of-distribution detection, and motivates a new perspective on data augmentation designed by re-combing the phase spectrum of the current image and the amplitude spectrum of the distracter image. That is, the generated samples force the CNN to pay more attention to the structured information from phase components and keep robust to the variation of the amplitude. Experiments on several image datasets indicate that the proposed method achieves state-of-the-art performances on multiple generalizations and calibration tasks, including adaptability for common corruptions and surface variations, out-of-distribution detection, and adversarial attack.