Improving the Robustness of Deep Neural Networks via Stability Training
This addresses robustness issues in computer vision models, which is crucial for real-world applications, but it is incremental as it builds on existing architectures like Inception.
The paper tackles the problem of output instability in deep neural networks caused by small input perturbations, and presents a stability training method that stabilizes networks against common image distortions, achieving robust state-of-the-art performance on tasks like near-duplicate detection and classification on noisy datasets.
In this paper we address the issue of output instability of deep neural networks: small perturbations in the visual input can significantly distort the feature embeddings and output of a neural network. Such instability affects many deep architectures with state-of-the-art performance on a wide range of computer vision tasks. We present a general stability training method to stabilize deep networks against small input distortions that result from various types of common image processing, such as compression, rescaling, and cropping. We validate our method by stabilizing the state-of-the-art Inception architecture against these types of distortions. In addition, we demonstrate that our stabilized model gives robust state-of-the-art performance on large-scale near-duplicate detection, similar-image ranking, and classification on noisy datasets.