Unsupervised anomaly detection refers to the discovery of unconventional images that are globally or locally different from the training set. Recently, reconstruction-based anomaly detection methods have made great progress. However, most of the existing methods take reconstructing the original image as the goal of latent feature learning. Due to lack of effective semantic guidance, latent features have intrinsic characteristics which retain redundant details of spatial structure. Such information is too general and cause over-expression problem. To solve this problem, in this paper, dual transformation-aware embeddings are coined which aims to achieve a stable model to learn high-level latent features in a self-supervised manner. To be more specific, the authors try to extract transformation-detectable feature embeddings for both structure and content views which explore the regular pattern under different transformations in normal situations. In addition, the relationship between the original feature and the transformed feature is established. Based on such relationship, the latent feature of generated image to predict transformation parameter is extracted. Then, a transformation-consistency regularization is proposed to constrain decoder to generate high-quality image with high-level consistency and achieve a more stable model. Experiments on MVTec-AD and CIFAR10 datasets prove the effectiveness and robustness of the proposed method.
Global Navigation Satellite System (GNSS) signals generate slant tropospheric delays when they pass through the atmosphere, which is recognized as the main source of error in many spatial geodetic applications. The zenith tropospheric delay (ZTD) derived from radio occultation data is of great significance to atmospheric research and meteorology and needs to be assessed in the use of precision positioning. Based on the atmPrf, sonPrf, and echPrf data from the Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) Data Analysis and Archiving Center (CDAAC) from 1 January to 31 December 2008 and 2012, we obtained the ZTDs of the radio occultation data (occZTD) and the corresponding radiosonde (sonZTD) and ECWMF data (echZTD). The ZTDs derived from ground-based global positioning system (GPS) observations from the International GNSS Service (IGS) were corrected to the lowest tangent point height of the matched radio occultation profile by the barometric height formula (gnsZTD). The statistical results show that the absolute values of the bias between occZTD and echZTD, sonZTD, or gnsZTD are less than 5 mm, and the standard deviations are approximately 20 mm or less, indicating that occZTD had significant accuracy in the GNSS positioning model even when the local spherical symmetry assumption error was introduced when the Abel inversion algorithm was used to obtain the refractive index profile of atmPrf. The effects of the horizontal/vertical matching resolution and the variation in the station height/latitude on the biases of occZTD and gnsZTD were analyzed. The results can be used to quantify the performance of radio occultation data for tropospheric delay error correction in dynamic high-precision positioning.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.