Distribution drift is an important issue for practical applications of machine learning (ML). In particular, in streaming ML, the data distribution may change over time, yielding the problem of ...concept drift, which affects the performance of learners trained with outdated data. In this article, we focus on supervised problems in an online nonstationary setting, introducing a novel learner-agnostic algorithm for drift adaptation, namely (), with the goal of performing efficient retraining of the learner when drift is detected. incrementally estimates the joint probability density of input and target for the incoming data and, as soon as drift is detected, retrains the learner using importance-weighted empirical risk minimization. The importance weights are computed for all the samples observed so far, employing the estimated densities, thus, using all available information efficiently. After presenting our approach, we provide a theoretical analysis in the abrupt drift setting. Finally, we present numerical simulations that illustrate how competes and often outperforms state-of-the-art stream learning techniques, including adaptive ensemble methods, on both synthetic and real-world data benchmarks.
In concept drift adaptation, we aim to design a blind or an informed strategy to update our best predictor for future data at each time point. However, existing informed drift adaptation methods need ...to wait for an entire batch of data to detect drift and then update the predictor (if drift is detected), which causes adaptation delay. To overcome the adaptation delay, we propose a sequentially updated statistic, called drift-gradient to quantify the increase of distributional discrepancy when every new instance arrives. Based on drift-gradient, a segment-based drift adaptation (SEGA) method is developed to online update our best predictor. Drift-gradient is defined on a segment in the training set. It can precisely quantify the increase of distributional discrepancy between the old segment and the newest segment when only one new instance is available at each time point. A lower value of drift-gradient on the old segment represents that the distribution of the new instance is closer to the distribution of the old segment. Based on the drift-gradient, SEGA retrains our best predictors with the segments that have the minimum drift-gradient when every new instance arrives. SEGA has been validated by extensive experiments on both synthetic and real-world, classification and regression data streams. The experimental results show that SEGA outperforms competitive blind and informed drift adaptation methods.
The recently emerged SARS-CoV-2 Omicron variant encodes 37 amino acid substitutions in the spike protein, 15 of which are in the receptor-binding domain (RBD), thereby raising concerns about the ...effectiveness of available vaccines and antibody-based therapeutics. Here we show that the Omicron RBD binds to human ACE2 with enhanced affinity, relative to the Wuhan-Hu-1 RBD, and binds to mouse ACE2. Marked reductions in neutralizing activity were observed against Omicron compared to the ancestral pseudovirus in plasma from convalescent individuals and from individuals who had been vaccinated against SARS-CoV-2, but this loss was less pronounced after a third dose of vaccine. Most monoclonal antibodies that are directed against the receptor-binding motif lost in vitro neutralizing activity against Omicron, with only 3 out of 29 monoclonal antibodies retaining unaltered potency, including the ACE2-mimicking S2K146 antibody
. Furthermore, a fraction of broadly neutralizing sarbecovirus monoclonal antibodies neutralized Omicron through recognition of antigenic sites outside the receptor-binding motif, including sotrovimab
, S2X259
and S2H97
. The magnitude of Omicron-mediated immune evasion marks a major antigenic shift in SARS-CoV-2. Broadly neutralizing monoclonal antibodies that recognize RBD epitopes that are conserved among SARS-CoV-2 variants and other sarbecoviruses may prove key to controlling the ongoing pandemic and future zoonotic spillovers.
Collision cross section (CCS) measurements resulting from ion mobility–mass spectrometry (IM-MS) experiments provide a promising orthogonal dimension of structural information in MS-based analytical ...separations. As with any molecular identifier, interlaboratory standardization must precede broad range integration into analytical workflows. In this study, we present a reference drift tube ion mobility mass spectrometer (DTIM-MS) where improvements on the measurement accuracy of experimental parameters influencing IM separations provide standardized drift tube, nitrogen CCS values (DTCCSN2) for over 120 unique ion species with the lowest measurement uncertainty to date. The reproducibility of these DTCCSN2 values are evaluated across three additional laboratories on a commercially available DTIM-MS instrument. The traditional stepped field CCS method performs with a relative standard deviation (RSD) of 0.29% for all ion species across the three additional laboratories. The calibrated single field CCS method, which is compatible with a wide range of chromatographic inlet systems, performs with an average, absolute bias of 0.54% to the standardized stepped field DTCCSN2 values on the reference system. The low RSD and biases observed in this interlaboratory study illustrate the potential of DTIM-MS for providing a molecular identifier for a broad range of discovery based analyses.
•A new void fraction correlation for a wide range of gas–liquid two phase flow.•Proposed correlation is based on drift flux model and is flow pattern independent.•Drift flux parameters are modeled as ...a function of several two phase flow variables.•Proposed correlation verified against a comprehensive data bank of 8255 data points.•Proposed correlation gives better accuracy compared to other correlations.
The main objective of this study is to present new equations for a flow pattern independent drift flux model based void fraction correlation applicable to gas–liquid two phase flow covering a wide range of fluid combinations and pipe diameters. Two separate sets of equations are proposed for drift flux model parameters namely, the distribution parameter (Co) and the drift velocity (Ugm). These equations for Co and Ugm are defined as a function of several two phase flow variables and are shown to be in agreement with the two phase flow physics. The underlying data base used for the performance verification of the proposed correlation consists of experimentally measured 8255 data points collected from more than 60 sources that consists of air–water, argon–water, natural gas–water, air–kerosene, air–glycerin, argon–acetone, argon–ethanol, argon–alcohol, refrigerants (R11, R12, R22, R134a, R114, R410A, R290 and R1234yf), steam–water and air–oil fluid combinations. It is shown that the proposed correlation successfully predicts the void fraction with desired accuracy for hydraulic pipe diameters in a range of 0.5–305mm (circular, annular and rectangular pipe geometries), pipe orientations in a range of -90°⩽θ⩽90°, liquid viscosity in a range of 0.0001–0.6Pa-s, system pressure in a range of 0.1–18.1MPa and two phase Reynolds number in a range of 10 to 5×106. Moreover, the accuracy of the proposed correlation is also compared with some of the existing top performing correlations based on drift flux and separated flow models. Based on this comparison, it is found that the proposed correlation consistently gives better performance over the entire range of the void fraction (0<α<1) and is recommended to predict void fraction without any reference to flow regime maps.
We present a brute-force approach to detect concept drift behind time sequence data. This approach, named Select-Starţ searches for start points of concept drift to minimize error. In other words, ...Select-Start searches for the start points of new concepts from the input sequence. Unlike many related works, Select-Start does not require a pre-specified error threshold to detect drift. This paper compares Select-Start with previous representative methods and clarifies its characteristics. The experimental results show that Select-Start is accurate for concept drift problems where the threshold changes slowly. However, existing methods are better at analyzing concept drift problems where the model behind data changes rapidly.
The Ion Velocity Meter (IVM) on NASA’s Ionospheric Connection Explorer (ICON) reports the in-situ ion density, ion temperature and 3-component ion drift velocity, retrieved from measurements by a ...retarding potential analyzer and an ion drift meter. ICON was launched during a deep solar minimum in late 2019, followed by a solar quiet (F10.7 < 80) period until September 2020. In order to quantify the uncertainties in the IVM’s drift velocity in a low plasma density environment, we compared IVM’s vertical drift velocity with eastward electric field (EEF) obtained from Swarm’s equatorial electrojet current measurements, the vertical drift from ground-based incoherent scatter radar (ISR) at Jicamarca Radio Observatory (JRO) and from Jicamarca Unattended Long-term studies of Ionosphere and Atmosphere (JULIA) coherent mode. The main results of this study show that (1) the vertical drift derived from Swarm’s EEF and ISR are in good agreement with the zonal electric field derived from JULIA’s vertical drift regardless of the F10.7 value. (2) The zonal electric field derived from IVM’s meridional drift is in good agreement with Swarm’s EEF in 2021, whereas the distribution is highly scattered in the deepest solar minimum in 2020. (3) An ad hoc IVM correction based on the 24-hour running mean of meridional drift can bring the IVM data into better agreement with Swarm and JULIA. An additional quality control based on O
+
fractional composition may be needed for some studies using IVM’s vertical drift. By using the same methodology presented in this work, future missions could calibrate their drift measurements to facilitate meaningful integration with ICON/IVM observations through the comparision with ground-based measurements.
Hemispheric asymmetries of the Vertical Total Electron Content (VTEC) were observed during the first recovery phase of the geomagnetic storm on September 7–8, 2017. These asymmetries occurred at the ...mid latitudes at two different local times simultaneously: In the European‐African sector (early morning), the storm time VTEC in the southern/northern hemisphere was higher/lower than the quiet time value, suggesting the southern/northern hemisphere entered the positive/negative phase (N−S+). In the East Asian‐Australian sector (afternoon), the storm time VTEC change was positive in the northern hemisphere, but negative in the southern hemisphere (N+S−). The electron density profiles from digisondes demonstrated that the asymmetries appeared in the F region density as well. The plasma drifts data from digisondes, the column‐integrated O/N2 ratio from GUVI onboard the TIMED satellite, and the detrended VTEC were utilized to study the drivers of the asymmetries. Traveling Ionospheric Disturbance (TID) signatures were identified in the digisonde drift and detrended VTEC data before the appearance of the asymmetry. The magnitude of TIDs was larger in the hemisphere where the negative phase occurred later. The storm time O/N2 ratio change was positive in Africa (S+) and negative in Europe (N−). However, the O/N2 measurements were not available in the East Asian‐Australian sector during the focused period. The hemispheric differences in the vertical drifts were also observed in both sectors. Therefore, the observed hemispheric asymmetries in both sectors are suggested to be due to the hemispheric asymmetries in the thermospheric composition change, vertical drift, and TID activity.
Key Points
Hemispheric asymmetries of the mid‐latitude ionosphere were observed during the first recovery phase of the September 7–8, 2017 storm
Hemispheric asymmetries were opposite over the European‐African and East Asian‐Australian sectors simultaneously
Their formation is likely due to the asymmetries of the thermospheric composition change, vertical plasma drift, and Traveling Ionospheric Disturbance activity
Organo-auxin (phenoxy) herbicides have found a place in weed control schemes for peanut, corn, small grains, sugarcane, turf, pasture and forage crops, and many other areas. It is the intent of this ...publication to clarify and disseminate the Florida Organo-Auxin Herbicide Rule to interested growers and applicators. Major revision by B. Bultemeier, J. A. Ferrell, and G. E. MacDonald; 4 pages. https://edis.ifas.ufl.edu/wg051
With the application of numerous services or software, process mining has attracted more and more attention. However, concept drift may occur during process mining due to the instability of the ...process. Sudden and gradual drifts are considered to be two basic modes of change, that may always appear in independent or nested forms. Although the existing methods have studied the detection of two basic modes, they do not consider the nesting of two change modes. We identify the change mode that sudden drifts and gradual drifts do not appear independently as nested drifts. The current drift detection methods can only detect the drift modes that occur independently, but not suitable for nested drift detection. To fill this gap, this paper proposes a business concept drift detection and localization framework called BRDDL (Behavior Replacement-based Drift Detection and Localization) which can not only detect independent drifts such as sudden drifts and gradual drifts, but also detect nested drifts. Firstly, we propose an integrated drift point detection and localization method which can report the location of change points and return the changed behaviors (activity relationship pairs). On this basis, we propose a behavior replacement method by updating the changed traces to restore an unchanged sub log. Then we compare the behaviors in the updated traces with those in the associated unchanged traces to judge the type of drifts. The effectiveness of the method is verified by simulation experiments on the synthetic log.