Search for: All records

Abnormal event detection with the lowest latency is an indispensable function for safety-critical systems, such as cyber defense systems. However, as systems become increasingly complicated, conventional sequential event detection methods become less effective, especially when we need to define indicator metrics from complicated data manually. Although Deep Neural Networks (DNNs) have been used to handle heterogeneous data, the theoretic assurability and explainability are still insufficient. This paper provides a holistic framework for the quickest and sequential detection of abnormalities and time-dependent abnormal events. We explore the latent space characteristics of zero-bias neural networks considering the classification boundaries and abnormalities. We then provide a novel method to convert zero-bias DNN classifiers into performance-assured binary abnormality detectors. Finally, we provide a sequential Quickest Detection (QD) scheme that provides the theoretically assured lowest abnormal event detection delay under false alarm constraints using the converted abnormality detector. We verify the effectiveness of the framework using real massive signal records in aviation communication systems and simulation. Codes and data are available at. 

Abstract The flow and transformation of warm, salty Atlantic‐origin water (AW) in the Arctic Ocean plays an important role in the global overturning circulation that helps regulate Earth's climate. The heat that it transports also impacts ice melt in different parts of the Arctic. This study uses data from a mooring array deployed across the shelf/slope of the Alaskan Beaufort Sea from 2002–2004 to investigate the flow of AW. A short‐lived “rebound jet” of AW on the upper continental slope regularly follows wind‐driven upwelling events. A total of 57 such events, lasting on average 3 days each, occurred over the 2 year period. As the easterly wind subsides, the rebound jet quickly spins up while the isopycnals continue to slump from their upwelled state. The strength of the jet is related to the cross‐slope isopycnal displacement, which in turn is dependent on the magnitude of the wind, in line with previous modeling. Seaward of the rebound jet, the offshore‐most mooring of the array measured the onshore branch of the AW boundary flowing eastward in the Canada Basin. However, the signature of the boundary current was only evident in the second year of the mooring timeseries. We suspect that this is due to the varying influence of the Beaufort Gyre in the two years, associated with a change in pattern of the wind stress curl that helps drive the gyre. 

Abstract Synoptic shipboard measurements, together with historical hydrographic data and satellite data, are used to elucidate the detailed structure of the Atlantic Water (AW) boundary current system in the southern Canada Basin and its connection to the upstream source of AW in the Chukchi Borderland. Nine high‐resolution occupations of a transect extending from the Beaufort shelf to the deep basin near 152°W, taken between 2003 and 2018, reveal that there are two branches of the AW boundary current that flow beneath and counter to the Beaufort Gyre. Each branch corresponds to a warm temperature core and transports comparable amounts of Fram Strait Branch Water between roughly 200–700 m depth, although they are characterized by a different temperature/salinity (T/S) structure. The mean volume flux of the combined branches is 0.87 ± 0.13 Sv. Using the historical hydrographic data, the two branches are tracked upstream by their temperature cores andT/Ssignatures. This sheds new light on how the AW negotiates the Chukchi Borderland and why two branches emerge from this region. Lastly, the propagation of warm temperature anomalies through the region is quantified and shown to be consistent with the deduced circulation scheme.