Statistically Robust Neural Network Classification
Despite their numerous successes, there are many scenarios where adversarial risk metrics do not provide an appropriate measure of robustness. For example, test-time perturbations may occur in a probabilistic manner rather than being generated by an explicit adversary, while the poor train--test generalization of adversarial metrics can limit their usage to simple problems. Motivated by this, we develop a probabilistic robust risk framework, the statistically robust risk (SRR), which considers pointwise corruption distributions, as opposed to worst-case adversaries. The SRR provides a distinct and complementary measure of robust performance, compared to natural and adversarial risk. We show that the SRR admits estimation and training schemes which are as simple and efficient as for the natural risk: these simply require noising the inputs, but with a principled derivation for exactly how and why this should be done. Furthermore, we demonstrate both theoretically and experimentally that it can provide superior generalization performance compared with adversarial risks, enabling application to high-dimensional datasets.
Code (1)
Tasks
ClassificationGeneral ClassificationSimilar Papers 제목 키워드 기반
Adversarial Consistency and the Uniqueness of the Adversarial Bayes Classifier
Minimizing an adversarial surrogate risk is a common technique for learning robust classifiers. Prior work showed that convex surrogate losses are not statistically consistent in the adversarial context -- or in other wo…
ClassificationAnalysis of the Effect of Unexpected Outliers in the Classification of Spectroscopy Data
Multi-class classification algorithms are very widely used, but we argue that they are not always ideal from a theoretical perspective, because they assume all classes are characterized by the data, whereas in many appli…
Binary ClassificationClassificationGeneral ClassificationMulti-class ClassificationCost-effective vibration analysis through data-backed pipeline optimisation
Vibration analysis is an active area of research, aimed, among other targets, at an accurate classification of machinery failure modes. This often leads to complex and convoluted signal processing pipeline designs, which…
I Am Big, You Are Little; I Am Right, You Are Wrong
Machine learning for image classification is an active and rapidly developing field. With the proliferation of classifiers of different sizes and different architectures, the problem of choosing the right model becomes m…
Image ClassificationStatistically Significant Concept-based Explanation of Image Classifiers via Model Knockoffs
A concept-based classifier can explain the decision process of a deep learning model by human-understandable concepts in image classification problems. However, sometimes concept-based explanations may cause false positi…
Deep Learningimage-classificationImage Classification