One-vs-Rest (OVR) classification aims to distinguish a single class of interest (COI) from other classes. The concept of novelty detection and robustness to dataset shift becomes crucial in OVR when the scope of the rest class is extended from the classes observed during training to unseen and possibly unrelated classes, a setting referred to as open set recognition (OSR). In this work, we propose a novel architecture, namely decoupling autoencoder (DAE), which provides a proven upper bound on the open space risk and minimizes open space risk via a dedicated training routine. Our method is benchmarked within three different scenarios, each isolating different aspects of OSR, namely plain classification, outlier detection, and dataset shift. The results conclusively show that DAE achieves robust performance across all three tasks. This level of cross-task robustness is not observed for any of the seven potent baselines from the OSR, OVR, outlier detection, and ensembling domain which, apart from ATA (Lübbering et al., From imbalanced classification to supervised outlier detection problems: adversarially trained auto encoders. In: Artificial neural networks and machine learning—ICANN 2020, 2020), tend to fail on either one of the tasks. Similar to DAE, ATA is based on autoencoders and facilitates the reconstruction error to predict the inlierness of a sample. However unlike DAE, it does not provide any uncertainty scores and therefore lacks rudimentary means of interpretation. Our adversarial robustness and local stability results further support DAE’s superiority in the OSR setting, emphasizing its applicability in safety-critical systems.
Background: The attainment of precise posterior occlusion alignment necessitates a deeper understanding of the clinical efficacy of aligner therapy. This study aims to determine whether the treatment goals defined in the virtual planning of aligner therapy are effectively implemented in clinical practice, with a particular focus on the influence of distalization distances on potential vertical side effects. Methods: In this retrospective, non-interventional investigation, a cohort of 20 individuals undergoing Invisalign® treatment was examined. Pre- and post-treatment maxillary clinical and ClinCheck® casts were superimposed utilizing a surface–surface matching algorithm on palatal folds, median palatine raphe, and unmoved teeth as the stable references. The effectivity of planned versus clinical movements was evaluated. Groupings were based on distalization distances, planned vertical movements, and Class II elastic prescription. Statistics were performed with a two-sample t-test and p-value < 0.05. Results: Clinically achieved distalization was significantly lower than virtually planned distalization, regardless of additional vertical movements, where a lack of implementation was contingent upon the extent of distalization, with no mitigating effects observed with the application of Class II elastics. Intriguingly, no adverse vertical side effects were noted; however, the intended intrusions or extrusions, as per the therapeutic plans, remained unattainable regardless of the magnitude of distalization. Conclusions: These findings underscore the imperative for future investigations to delve deeper into the intricacies surrounding translational mesio-distal and vertical movements, thereby enhancing predictability within orthodontic practice. To facilitate successful clinical implementation of vertical and translational movements via aligners, the incorporation of sliders emerges as a promising strategy for bolstering anchorage reinforcement.