The Effect of Optimization Methods on the Robustness of Out-of-Distribution Detection Approaches
This work addresses the unreliability of OOD detection in critical domains, but it is incremental as it focuses on optimization sensitivity rather than a new detection method.
The paper investigates how optimization methods affect the robustness of out-of-distribution detection approaches in deep neural networks, finding that some optimization methods provide better solutions and proposing a robustness score for comparison.
Deep neural networks (DNNs) have become the de facto learning mechanism in different domains. Their tendency to perform unreliably on out-of-distribution (OOD) inputs hinders their adoption in critical domains. Several approaches have been proposed for detecting OOD inputs. However, existing approaches still lack robustness. In this paper, we shed light on the robustness of OOD detection (OODD) approaches by revealing the important role of optimization methods. We show that OODD approaches are sensitive to the type of optimization method used during training deep models. Optimization methods can provide different solutions to a non-convex problem and so these solutions may or may not satisfy the assumptions (e.g., distributions of deep features) made by OODD approaches. Furthermore, we propose a robustness score that takes into account the role of optimization methods. This provides a sound way to compare OODD approaches. In addition to comparing several OODD approaches using our proposed robustness score, we demonstrate that some optimization methods provide better solutions for OODD approaches.