Fourier Sensitivity

Fourier sensitivity analyzes how sensitive models, particularly deep neural networks (DNNs), are to different frequency components in their input data. Current research focuses on measuring this sensitivity, developing regularization techniques to mitigate undesirable frequency biases, and leveraging frequency analysis for tasks like detecting adversarial attacks in DNNs. Understanding and controlling Fourier sensitivity is crucial for improving model robustness, accuracy, and trustworthiness across various applications, including computer vision and medical imaging.