Drift Model Research Articles

Analytical Solutions of the Transport Equation for a Drift Model in an Unsteady Flow with Discontinuous Parameters

Analytical Solutions of the Transport Equation for a Drift Model in an Unsteady Flow with Discontinuous Parameters

Ross–Weddell Dipole Critical for Antarctic Sea Ice Predictability in MPI–ESM–HR

We use hindcasts from a state-of-the-art decadal climate prediction system initialized between 1979 and 2017 to explore the predictability of the Antarctic dipole—that is, the seesaw between sea ice cover in the Weddell and Ross Seas, and discuss its implications for Antarctic sea ice predictability. Our results indicate low forecast skills for the Antarctic dipole in the first hindcast year, with a strong relaxation of March values toward the climatology contrasting with an overestimation of anomalies in September, which we interpret as being linked to a predominance of local drift processes over initialized large-scale dynamics. Forecast skills for the Antarctic dipole and total Antarctic sea ice extent are uncorrelated. Limited predictability of the Antarctic dipole is also found under preconditioning around strong warm and strong cold events of the El Niño-Southern Oscillation. Initialization timing and model drift are reported as potential explanations for the poor predictive skills identified.

Study of estimated model of drift through real ship

Study of estimated model of drift through real ship

Anticipative Bayesian classification for data streams with verification latency

Most of the existing adaptive classification algorithms in non-stationary data streams require recent labelled data for their updates. Such recent labels are often missing. For stream classification under verification latency only few approaches exist. Most of them assume clustered data or homogeneous drift in all features, which limits their applicability. We address this by proposing Anticipative Bayesian stream Classifier (ABClass), an approach that is capable of integrating and automatically selecting from different components. In its Bayesian classification framework, ABClass combines density estimation techniques, extended to extrapolate drift patterns over time, with unsupervised parameter tuning and unsupervised model selection. ABClass allows for multivariate density estimation and extrapolation techniques. In this work, we assume conditional independence between features given the class label for modelling feature-specific drift patterns. ABClass is generative and can also be used for explaining and visualising concept drift patterns. It is generic, making it easy to include further types of drift models, both for the class-conditional feature distribution and for the class prior distribution. The experimental evaluation on several real-world data streams shows its competitiveness compared to other state-of-the-art approaches. ABClass is in most cases ten- to hundred-times faster than its competitors, both for model fitting and for prediction.

Research on the Drift Prediction of Marine Floating Debris: A Case Study of the South China Sea Maritime Drift Experiment

Annually, hundreds of individuals tragically lose their lives at sea due to shipwrecks or aircraft accidents. For search and rescue personnel, the task of locating the debris of a downed aircraft in the vastness of the ocean presents a formidable challenge. A primary task these teams face is determining the search area, which is a critical step in the rescue operation. The movement of aircraft wreckage on the ocean surface is extremely complex, influenced by the combined effects of surface winds, waves, and currents. Establishing an appropriate drift motion prediction model is instrumental in accurately determining the search area for the wreckage. This article initially conducts maritime drift observation experiments on wreckage, and based on the results of these experiments, analyzes the drift characteristics and patterns of the debris. Subsequently, employing a wealth of observational experimental data, three types of drift prediction models for the wreckage are established using the least squares method. These models include the AP98 model, the dynamics model, and an improved model. In conclusion, the effectiveness and accuracy of the three models is evaluated and analyzed using Monte Carlo techniques. The results indicate that the probability of positive crosswind leeway (CWL) is 47.4%, while the probability of negative crosswind leeway (CWL) is 52.6%. The jibing frequency is 7.7% per hour, and the maximum leeway divergence angle observed is 40.4 degrees. Among the three drift prediction models, the refined AP98 drift model demonstrates the highest forecasting precision. The findings of this study offer a more accurate drift prediction model for the search of an aircraft lost at sea. These results hold significant guiding importance for maritime search and rescue operations in the South China Sea.

Multiple correlation filters with gaussian constraint for fast online tracking

The correlation filter based online tracking methods usually can achieve high real-time performance due to the leverage of the well-known FFT. However, they are also apt to generate the “corrupted training samples” in scenarios with complex background, which will trigger the model drift and deteriorate the tracking accuracy rapidly. The existing methods usually consider this problem from certain aspect and none of them has mined the potential of combining multiple formulas. In this paper, we propose the Output Constraint Transfer - Discriminative Scale Space Tracking (OCT-DSST) algorithm, which has taken full consideration of multiple channel feature, multiple filters, kernel trick, memory with incremental learning, and the self-supervision mechanism. We re-formulate the online tracking process by combining all formulas above in a unified framework. The so obtained adaptive learning rate can better exploit the feedback information coming from the intermediate tracking results, and effectively mitigate the corrupted sample problem. The experimental results on the OTB-50/100 and the VOT2016 datasets reveal that the proposed method is comparative to most state-of-the-arts algorithms, and can increase the accuracy by 2% and the success rate by 1.7%, compared to the traditional DSST method.

Susceptibility of AutoML mortality prediction algorithms to model drift caused by the COVID pandemic

BackgroundConcept drift and covariate shift lead to a degradation of machine learning (ML) models. The objective of our study was to characterize sudden data drift as caused by the COVID pandemic. Furthermore, we investigated the suitability of certain methods in model training to prevent model degradation caused by data drift.MethodsWe trained different ML models with the H2O AutoML method on a dataset comprising 102,666 cases of surgical patients collected in the years 2014–2019 to predict postoperative mortality using preoperatively available data. Models applied were Generalized Linear Model with regularization, Default Random Forest, Gradient Boosting Machine, eXtreme Gradient Boosting, Deep Learning and Stacked Ensembles comprising all base models. Further, we modified the original models by applying three different methods when training on the original pre-pandemic dataset: (1) we weighted older data weaker, (2) used only the most recent data for model training and (3) performed a z-transformation of the numerical input parameters. Afterwards, we tested model performance on a pre-pandemic and an in-pandemic data set not used in the training process, and analysed common features.ResultsThe models produced showed excellent areas under receiver-operating characteristic and acceptable precision-recall curves when tested on a dataset from January-March 2020, but significant degradation when tested on a dataset collected in the first wave of the COVID pandemic from April-May 2020. When comparing the probability distributions of the input parameters, significant differences between pre-pandemic and in-pandemic data were found. The endpoint of our models, in-hospital mortality after surgery, did not differ significantly between pre- and in-pandemic data and was about 1% in each case. However, the models varied considerably in the composition of their input parameters. None of our applied modifications prevented a loss of performance, although very different models emerged from it, using a large variety of parameters.ConclusionsOur results show that none of our tested easy-to-implement measures in model training can prevent deterioration in the case of sudden external events. Therefore, we conclude that, in the presence of concept drift and covariate shift, close monitoring and critical review of model predictions are necessary.

LoRaSync: Energy efficient synchronization for scalable LoRaWAN

AbstractLow‐power wide‐area networks connect a large number of battery‐powered wireless devices over long distances. Among them, long range wide area networks (LoRaWAN) implement a pure ALOHA medium access scheme to save device energy by minimizing the radio usage. However, frame collisions restrain the network scalability when the traffic load increases. In this context, synchronization can be used to exploit the available bandwidth more efficiently by controlling the timing of frame transmissions and reducing the collision probability. Such strategy allows to increase the network throughput at the cost of an extra energy demand due to the inherent overhead. In that, the entailed network scenario is still supposed to address low power applications. Therefore this article timely presents LoRaSync, an energy‐efficient synchronization scheme designed for LoRa networks of any size. An accurate clock drift model was established based on measurements made on real cheap devices, and leveraged to support the design of LoRaSync. Our mechanism has been used to evaluate the same ALOHA‐based random access but on a time‐slotted basis, thus increasing the maximum achievable throughput compared to the legacy access. Throughput and energy efficiency models are established to evaluate the performances of a LoRaSync‐operated network. These models are validated with a simulation environment mimicking large‐scale deployments, and then used to determine the most energy efficient slot size for any traffic load. As a final proof of concept, LoRaSync has been implemented and tested on a LoRa testbed to demonstrate the feasibility of our solution on real hardware.

Statistical forecasting of U.S. and Central African Republic net migration

Immigration is a very important link in the current international society. This paper will study and predict the net immigration of the United States and the Central African Republic through two different models- drift model and ARIMA model, and to further explore the trends and influencing factors of migration between these countries. The results show that from 1960 to 2021, net migration from the United States and the Central African Republic showed very different trends. The United States, as a developed country, attracts a large number of immigrants from all over the world, while the Central African Republic, as a developing country, the flow of immigrants is mainly affected by economic, political and social factors in the region. Therefore, it can be seen that developing countries and developed countries have different impacts on the number of immigrants. This study provides a basis for further understanding of population migration and net migration of United States and Central African Republic.

Respiratory particles: from analytical estimates to disease transmission

Respiratory particles containing infectious pathogens are responsible for a large number of diseases. To define health politics and save lives, it is important to study their transmission mechanisms, namely the path of particles once expelled. This path depends on several driving factors as intrinsic properties of particles, environmental aspects and morphology of the scenario. Following physical arguments and taking into account the results of experimental works, we consider a mathematical drift model for the mixture composed by two phases: air and particles. The relative motion between the two phases is described by a kinematic constitutive relation. We prove the stability of the model for fixed times and establish an a priori estimate for the total number of infectious particles. The upper bound of this estimate exhibits sound physical dependencies on the driving factors, in agreement with the experimental literature and mounting epidemiological evidences. Namely, we establish that the amount of particles expelled and their emission rate can explain why some people are superspreaders. Several numerical simulations illustrate the theoretical results.

The modeling and analysis of operating characteristics and temperature drift of deflector jet servo valve under the wide temperature range

The Deflector Jet Servo Valve (DJSV) holds a crucial role within the electro-hydraulic servo system, as its performance directly influences the equipment reliability. DJSV operates within a wide temperature range, and the temperature fluctuations result in the magnetomotive force variations of the permanent magnets and changes in the working air gaps, which influence the operating characteristics of the torque motor and contribute to temperature drift. Consequently, a theoretical model is proposed to comprehensively analyze the impact of these factors on the operating characteristics and temperature drift of the DJSV. The results indicate that as the temperature increases, the magnetic flux of working air gap decreases, while the temperature drift increases. Additionally, among the five initial air gap errors discussed, a centrosymmetric error of less than 0.02 l can significantly reduce the temperature drift. The proposed modeling and analysis method of temperature drift can be used to guide the design and analysis of DJSV in wide temperature range.

Numerical simulation validating of a duck-mouth type marine debris collection device's collection mechanism

Marine debris (MD) poses a significant threat to global coastal ecosystems around the world, necessitating effective strategies for its collection and removal. As a new type of fixed collection device, the duck-mouth type marine debris collection device (MDCD) consists primarily of two components: a central collection system and floating fences which are positioned at a certain angle and open towards the outer bay located on both sides of it. This paper aims to establish a coupling drift model based on hydrodynamic model to study the performance of duck-mouth type MDCD, verify its effectiveness. Before simulation, the model was fully validated. The results demonstrate that wind has the greatest influence on MD movement, as the direction of the wind directly determines the movement direction of the debris. It was observed that only under onshore wind conditions did the MD move towards the bay when the duck-mouth type MDCD can effectively collect MD, which moves along the barrier net towards the central trash bin and eventually be fully collected.

Role of Improved Ocean Initial State in the Seasonal Prediction of Indian Summer Monsoon: A Case Study

This case study has made an effort to show the impact of improved ocean initial conditions (ICs) in a coupled forecast system (CFSv2) simulation on the seasonal prediction of Indian summer monsoon rainfall (ISMR). CFSv2 is used as an operational dynamical model for the seasonal prediction of ISMR. Here, we show an improved ISMR skill by initializing the ocean component of CFSv2 using new improved ocean ICs based on Global Ocean Data Assimilation System (GODAS) analysis. This new analysis is better than the NCEP GODAS, which uses the earlier-generation ocean model MOM4p0d and assimilates observed temperature and synthetic salinity using the 3DVar assimilation scheme. However, the new, improved GODAS analysis uses the MOM4p1 ocean model and assimilates observed salinity instead of synthetic salinity. We performed twin sets of nearly identical model experiments differing only in their ICs, with one set using NCEP ICs and the other using the new ICs (NIC). The NIC experiment consistently shows better El Niño–Southern Oscillation prediction skill than the NCEP IC experiment. This advancement leads to improvement in the ISMR skill. We found that the substantial improvements in both oceanic and atmospheric variables in a coupled feedback system contributed to the improved ISMR skills. The enhanced ISMR skill score of the NIC experiment might be the result of improved teleconnections, better depiction of large-scale monsoon circulations, and reduced model drift.

The concept of service model fidelity in Talking Therapies

Abstract The concept of Service Model Fidelity is considered as a parallel process to Treatment Fidelity in evidence-based psychological therapies. NHS Talking Therapies (formerly IAPT) aimed to increase access to an expanded, upskilled workforce on a national scale. This included systematic training, supervision and front-line service delivery, emphasising treatment fidelity to evidence-based interventions. A further feature of NHS Talking Therapies was modernising and restructuring of the health system that housed these trained practitioners. The term ‘service model fidelity’ (Cromarty, 2016) was coined to emphasise service modernisation aspects as a distinct entity. A definition of the latter is included. Examples of service model fidelity and of service model drift, are outlined to distinguish these from therapist drift. This links to past literature recommending changes in traditional mental health service design and emergent evidence from NHS Talking Therapies. The latter examines publicly available data identifying characteristics of service design, which appear to be predictors of enhanced clinical outcome. Challenges in modernising health systems are discussed and conclusions are made highlighting the crucial role of service model when delivering evidence-based therapies. Suggestions for further research into service configuration to improve experiences of service users are considered. This includes ongoing exploration of service design being more than a qualitative feature, and increasingly appearing as a key factor in enhanced clinical outcome. Key learning aims (1) To identify service model fidelity as separate entity to treatment fidelity. (2) To provide a clear definition of service model fidelity. (3) To delineate therapist drift from service drift. (4) To further examine the role of service model in delivering evidence-based interventions.

Active Learning with Pre-trained Language Models for Named Entity Recognition in Requirements Engineering

The ever-increasing number of textual requirements and the handling of these consumes significant time in product design. Therefore, automating requirements analysis with natural language processing is critical to increase efficiency. Pre-trained language models have proven useful for requirements analysis. However, to deploy pre-trained language models, a fully labeled dataset must first be created to finetune language models towards the requirements domain. The creation of a fully labeled dataset, especially for named entity recognition (NER) applications, poses a major challenge as it entails high manual labeling effort by requirements engineers. Furthermore, in an ever-changing requirements environment, the manual labeling process must be repeated to avoid model drift, adding additional workload. Recent advances of pre-trained language models combined with active learning can significantly reduce the manual labeling effort. Thus, we present a novel approach for leveraging active learning with pre-trained language models for NER. We demonstrate the efficacy of this combination in an experiment, where we are able to reduce the manual requirements labeling effort by 74% while even improving model performance. We argue that by reducing the manual effort, active learning can shorten the ramp-up time of language model deployment for requirements engineering and enable industrial adoption.

The quantitative paradigm and the nature of the human mind. The replication crisis as an epistemological crisis of quantitative psychology in view of the ontic nature of the psyche.

Many suggestions for dealing with the so-called replication crisis in psychology revolve around the idea that better and more complex statistical-mathematical tools or stricter procedures are required in order to obtain reliable findings and prevent cheating or publication biases. While these aspects may play an exacerbating role, we interpret the replication crisis primarily as an epistemological crisis in psychology caused by an inadequate fit between the ontic nature of the psyche and the quantitative approach. On the basis of the philosophers of science Karl Popper, Thomas Kuhn, and Imre Lakatos we suggest that the replication crisis is therefore a symptom of a fundamental problem in psychology, but at the same time it is also an opportunity to advance psychology as a science. In a first step, against the background of Popper's Critical Rationalism, the replication crisis is interpreted as an opportunity to eliminate inaccurate theories from the pool of theories and to correct problematic developments. Continuing this line of thought, in an interpretation along the lines of Thomas Kuhn, the replication crisis might signify a model drift or even model crisis, thus possibly heralding a new paradigm in psychology. The reasons for this are located in the structure of academic psychology on the basis of Lakatos's assumption about how sciences operate. Accordingly, one hard core that lies at the very basis of psychology may be found in the assumption that the human psyche can and is to be understood in quantitative terms. For this to be possible, the ontic structure of the psyche, i.e., its very nature, must also in some way be quantitatively constituted. Hence, the replication crisis suggests that the ontic structure of the psyche in some way (also) contains a non-quantitative dimension that can only be grasped incompletely or fragmentarily using quantitative research methods. Fluctuating and inconsistent results in psychology could therefore also be the expression of a mismatch between the ontic level of the object of investigation and the epistemic level of the investigation.

Drift Detection and Model Update using Unsupervised AutoML in IoT

This paper addresses the challenges of concept drift on the Internet of Things (IoT) environments and evaluates a machine-learning model's performance under varying data drift conditions using unsupervised Automatic Machine Learning (AutoML) anomaly detection techniques. By implementing a dynamic learning framework and employing advanced analytics, the study showcases the resilience of the proposed methodology against evolving data patterns. The results demonstrate the model's robust predictive capabilities, even in high drift scenarios, underscoring the importance of adaptive models in maintaining effective IoT security measures. The achieved improvement percentages can reach 46% for the F1 score.

Tackling data and model drift in AI: Strategies for maintaining accuracy during ML model inference

In machine learning (ML) and artificial intelligence (AI), model accuracy over time is very important, particularly in dynamic environments where data and relationships change. Data and model drift pose challenging issues that this paper seeks to explore: shifts in input data distributions or underlying model structures that continuously degrade predictive performance. It analyzes different drift types in-depth, including covariate, prior probability, concept drift for dasta, parameters, hyperparameter, and algorithmic model drift. Key causes, ranging from environmental changes to evolving data sources and overfitting, contribute to decreased model reliability. The article also discusses practical strategies for detecting and mitigating Drift, such as regular monitoring, statistical tests, and performance tracking, alongside solutions like automated recalibration, ensemble methods, and online learning models to enhance adaptability. Furthermore, the importance of feedback loops and computerized systems in handling Drift is emphasized, with real-world case studies illustrating drift impacts in financial and healthcare applications. Finally, future AI system drift management will be highlighted from emerging directions such as AI-based drift prediction, transfer learning, and robust model design.

Modeling and Forecasting of Spinach Production in Bangladesh

Background: In Bangladesh, spinach (Spinachia oleracea L.) is most frequently known as "Palong shak" or Bengal Spinach. Spinach is one of the most prominent vegetable crops grown around the earth. It is a quick-growing annual plant that only lives for one year. Although spinach can be cultivated at any time throughout the year, Bangladesh's production of spinach on a per-unit basis is very low in comparison to that of other advanced nations.
 Objective: This study explores the impacts of area, price, and weather parameter adaptability on the local production of spinach in Bangladesh employing annual data from time series covering 60 years.
 Materials and Methods: This investigation seeks to comprehend the linear relationship between spinach production, commercial price, geographic location, and meteorological characteristics in Bangladesh by applying stepwise regression analysis to find the most suitable model of Spinach production. In order to evaluate how well the multiple linear regression model fits the data, several diagnostic charts were constructed using the R programming language. To forecast the Spinach production, an autoregressive integrated moving average (ARIMA) model was applied.
 Results: According to the findings of the study, there was a positive correlation between spinach output and total harvest area under spinach cultivation. There is weak positive correlation between production and annual mean temperature and a weak negative correlation exists between production and annual mean rainfall. Also, production is perfectly positively correlated with price. Regression analysis revealed that the variables sales price and annual mean rainfall are the best predictors of spinach output. Every unit increase in the price of spinach is increases an additional 5.98 times the average of spinach being produced. An increase of 1 millimeter in annual mean rainfall (AMR) decreases 0.001 times spinach production while other predictors remain fixed. Also, ARIMA (1, 1, 0) with drift model forecasts for next fifteen years that the Spinach production is expected to increase considerably by 75000 tons.
 Conclusion: Finally, the study found that farmers' perspective on spinach cultivation could potentially experience a positive upward shift which is expected to last for the next fifteen years.

Forecast the house price index for California using different forecasting methods

Forecasting house price index is a useful and classic problem in real estate and investment fields. Predicting house price index in a region not only helps investors make sensible decisions but also aids the government in promulgating policy. This paper will use some simple forecasting models (mean model, nave model, drift model, linear model and ARIMA model) in forecast test part and by seeing the average value of their residuals and checking whether the distribution of the residuals approximates the normal distribution, select the one with the highest accuracy among them for the final prediction. Multiple linear regression is also used to find if there is relationship between predicted data and possible influencing factors (such as income, unemployment rate and population) and then use the factors that have strong correlation with predicted data to optimize our forecasts and provide a more accurate prediction for the house price index in California in the next few years.