Uncertainty-based Approach Research Articles

Adapting East and Southern Africa’s livestock to climate change: a decision making under deep uncertainty-based approach for effective actions

ABSTRACT Livestock farmers are increasingly challenged to adapt to the impacts of climate change, necessitating the selection of adaptation strategies to effectively mitigate risks and protect livelihoods. This paper introduces a framework designed specifically for guiding the selection of context-specific adaptation options in the Eastern and Southern Africa region. The framework builds on a decision tree that incorporates changes within a management system or switching to another one, enabling a nuanced evaluation of adaptation options. Driven repetitively under different scenarios of climate changes and/or climate models, the frequencies of selecting different adaptation measures vary across livestock value chains, climate zones, and systems. Responding to the evolution of the climate system, these frequencies evolve over time, affecting the selection. For instance, agroforestry emerges as an increasingly suitable option for cattle and, to a lesser extent, for goats due to the projected rise in moderate heat stress periods, particularly in tropical climates. Conversely, this frequency decreases for sheep, more susceptible to heat stress, beyond the effect of agroforestry. This framework resolves the need for more context – and time-specific decisions on adaptation. This decision tree-based framework serves as a robust decision-making tool to steer the livestock sector toward effective climate change adaptation.

Tissue segmentation for traumatic brain injury based on multimodal MRI image fusion-semantic segmentation

Accurate segmentation of traumatic brain injury (TBI) has great significance for physicians to diagnose and assess a patient’s condition. The utilization of multimodal information plays a critical role in TBI segmentation. However, most of the existing methods mainly focus on direct extraction and selection of deep semantic features, whereas in this paper, we use image fusion as an auxiliary task for feature learning based on multimodal feature extraction to achieve more sufficient fusion of multimodal features. Therefore, we design a multimodal image fusion-semantic segmentation based framework. The proposed approach mainly consists of a semantic encoder module, a semantic segmentation module and an image fusion module. The semantic encoder compresses the input image into a smaller feature space to extract semantic features. The semantic segmentation module utilizes both the detailed information extracted by the encoder and the semantic information of high-level features extracted from the semantic segmentation module to generate the segmentation results. The image fusion module fuses semantic feature information from different modalities as an auxiliary task to semantic segmentation. Furthermore, to enhance the model’s performance even further, an uncertainty-based approach is employed, which dynamically adjusts the loss weights for the image fusion task and the semantic segmentation task during the model training process. The proposed method is evaluated on a private dataset, and compared with other widely recognized methods. It demonstrates outstanding performance in both Dice score and Recall metrics.

Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning

BackgroundRadiotherapy is a core treatment modality for oropharyngeal cancer (OPC), where the primary gross tumor volume (GTVp) is manually segmented with high interobserver variability. This calls for reliable and trustworthy automated tools in clinician workflow. Therefore, accurate uncertainty quantification and its downstream utilization is critical.MethodsHere we propose uncertainty-aware deep learning for OPC GTVp segmentation, and illustrate the utility of uncertainty in multiple applications. We examine two Bayesian deep learning (BDL) models and eight uncertainty measures, and utilize a large multi-institute dataset of 292 PET/CT scans to systematically analyze our approach.ResultsWe show that our uncertainty-based approach accurately predicts the quality of the deep learning segmentation in 86.6% of cases, identifies low performance cases for semi-automated correction, and visualizes regions of the scans where the segmentations likely fail.ConclusionsOur BDL-based analysis provides a first-step towards more widespread implementation of uncertainty quantification in OPC GTVp segmentation.

Water security and agricultural systems in the Galapagos Islands: vulnerabilities under uncertain future climate and land use pathways

The Galapagos Islands, a hotspot of ecological richness, face challenging climatic and development conditions which undermine regional water security. Yet, the way by which these conditions may change in the future is highly uncertain. In this study, we applied for the first time an uncertainty-based approach in the Galapagos Islands to understand thresholds and potential scenarios of risks to water security of agricultural catchments in the Galapagos Island. We applied a water systems model to the agricultural catchments as well as climate and land use scenarios to estimate physical changes in water availability and implications for drought and extreme flow thresholds. Our results highlight the key role of baseflow and its important sensitivity to precipitation changes in determining water security states in the Islands. In fact, a decrease in just about 25% of total water flows, from historical conditions, in the Islands would drive drought conditions resembling those of the emergency state of 2016. We also note that under a land use scenario which promotes sustainable practices, the robustness of the Islands to climate variations increases. Our study then provides the basis for an application of uncertainty-based approaches to enhance resilience of the agricultural water systems as well as other systems. Our results also emphasize the need to design flexible and comprehensive policies in the water, agricultural and interconnected sectors which consider the interlinkages of climate with other social and economic variables.

Anomaly Pixel Detection via Dual-Branch Uncertainty Metrics

Anomaly detection aims to detect abnormal samples from normal data. Existing methods mainly use synthesis-based approach and uncertainty-based approach. Moreover, the above-mentioned methods need to re-train part/all of the deep network, which limits its application scenarios especially real-time scenarios. We propose a novel retraining-free anomaly detection method, which includes a dual-branch anomaly detection mechanism to detect abnormal pixels from two views. In the pseudo-supervised branch, we use the Copy post method to transform the unsupervised anomaly detection task into a supervised binary detection task. In the prototype detection branch, we use prototypes to improve the accuracy of abnormal pixel detection. The anomaly results of the two branches are fused by means of soft voting to produce anomaly detection results with higher recall. Our experiments show that our method significantly outperforms other methods and achieves excellent results on different datasets, which also shows that our method has good generalization ability. Therefore, our proposed method can effectively improve the effectiveness and safety of anomaly detection in real-time scenarios, such as autonomous driving.

Rank-Aware Negative Training for Semi-Supervised Text Classification

Abstract Semi-supervised text classification-based paradigms (SSTC) typically employ the spirit of self-training. The key idea is to train a deep classifier on limited labeled texts and then iteratively predict the unlabeled texts as their pseudo-labels for further training. However, the performance is largely affected by the accuracy of pseudo-labels, which may not be significant in real-world scenarios. This paper presents a Rank-aware Negative Training (RNT) framework to address SSTC in learning with noisy label settings. To alleviate the noisy information, we adapt a reasoning with uncertainty-based approach to rank the unlabeled texts based on the evidential support received from the labeled texts. Moreover, we propose the use of negative training to train RNT based on the concept that “the input instance does not belong to the complementary label”. A complementary label is randomly selected from all labels except the label on-target. Intuitively, the probability of a true label serving as a complementary label is low and thus provides less noisy information during the training, resulting in better performance on the test data. Finally, we evaluate the proposed solution on various text classification benchmark datasets. Our extensive experiments show that it consistently overcomes the state-of-the-art alternatives in most scenarios and achieves competitive performance in the others. The code of RNT is publicly available on GitHub.

Simulating high-entropy alloys at finite temperatures: An uncertainty-based approach

A general method is presented for modeling high-entropy alloys as ensembles of randomly sampled, ordered configurations on a given lattice. Statistical mechanics is applied post hoc to derive the ensemble properties as a function of composition and temperature, including the free energy of mixing and local structure. Random sampling is employed to address the high computational costs needed to model alloys with a large number of components. Doing so also provides rigorous convergence criteria, including the quantification of noise due to random sampling, and an estimation of the number of additional samples required to lower this noise to the desired levels. Binary to five-component alloys of the group-IV chalcogenides are used as case examples, for which the predicted miscibility shows excellent agreement with experiment. This method is well-suited for calculating the configurational thermodynamics, local structure, and ensemble properties of complex alloys, and it is attractive for materials with temperature-dependent, short-range order.

UR-CarA-Net: A Cascaded Framework With Uncertainty Regularization for Automated Segmentation of Carotid Arteries on Black Blood MR Images

We present a fully automated method for carotid artery (CA) outer wall segmentation in black blood MRI using partially annotated data and compare it to the state-of-the-art reference model. Our model was trained and tested on multicentric data of patients (106 and 23 patients, respectively) with a carotid plaque and was validated on different MR sequences (24 patients) as well as data that were acquired with MRI systems of a different vendor (34 patients). A 3D nnU-Net was trained on pre-contrast T1w turbo spin echo (TSE) MR images. A CA centerline sliding window approach was chosen to refine the nnU-Net segmentation using an additionally trained 2D U-Net to increase agreement with manual annotations. To improve segmentation performance in areas with semantically and visually challenging voxels, Monte-Carlo dropout was used. To increase generalizability, data were augmented with intensity transformations. Our method achieves state-of-the-art results yielding a Dice similarity coefficient (DSC) of 91.7% (interquartile range (IQR) 3.3%) and volumetric intraclass correlation (ICC) with ground truth of 0.90 on the development domain data and a DSC of 91.1% (IQR 7.2%) and volumetric ICC with ground truth of 0.83 on the external domain data outperforming top-ranked methods for open-source CA segmentation. The uncertainty-based approach increases the interpretability of the proposed method by providing an uncertainty map together with the segmentation.

Uncertainty factor in contemporary communities of the Russian Arctic: methodological approaches to research

The article introduces readers to work on a new research project “Overcoming Uncertainty in Interaction with the Physical and Social Environment in the Russian Arctic” by describing its conceptual and theoretical approach and premises as well as the steps of designing it. It opens with an analysis of the concept of risk, which is currently popular in the Arctic Social Studies and which is sometimes claimed to be the central concept in the field. It proceeds by describing the new approach, which is based on the concept of uncertainty, both natural and social, which constantly accompanies human life in the Far North. The inhabitants of the North know from experience that uncertainty cannot be ruled out, but one can be constantly prepared for it by planning one’s behavior with this factor in mind. Furthermore, one can suggest that the effects of ecological, economic and social changes are experienced by these inhabitants first and foremost as shifts in the level and kind of uncertainty. Therefore, the concept of uncertainty can replace the concept of risk in the studies of the effects of change on northern communities; in fact, “uncertainty” is preferable because its conceived nature is obvious, while in many risk studies, risk is still perceived as something objective. The main questions that a study of uncertainty can answer are: what is the role of natural and social uncertainty in the everyday life of the permanent population of the Arctic? What are the action strategies of the people in the face of uncertainty, including discursive strategies to explain the growing uncertainty? What is the role of state authorities, industrial companies, local organizations, urban and rural communities in overcoming the negative consequences of natural and social uncertainties? How are social connections and networks involved in dealing with situations of uncertainty? The paper demonstrates how these questions can be answered on the basis of anthropological fieldwork. It also uses examples from the Komi Republic (Vorkuta), Yakutia (Chersky settlement), and Magadan area to demonstrate the sort of problems the uncertainty-based approach can contribute to solving. A review of the existing literature in Russian and English is presented in order to give the reader useful insights into the field of uncertainty studies and outline the main directions of research that a study of uncertainty can take

Screening of future carbon storage sites – selecting the best spots

Subsurface carbon storage can occur in depleted oil and gas fields, in water-wet structures, or in open aquifers. All three types of storage sites present advantages and inconveniences, which will be reviewed in this talk. The selection of future sites for carbon storage balances storage capacity (how much CO2 can be stored), injectivity (how efficiently or fast CO2 can be stored), and containment risk (how safely CO2 can be stored). We present a rigorous uncertainty-based approach involving estimates of pore volume, pressure and temperature conditions and resulting fluid properties, and sealing and containment behaviour, to highlight areas with best potential for safe and effective carbon storage.

City-wide model-based analysis of heat recovery from wastewater using an uncertainty-based approach

Around 90% of the energy requirement for urban water systems management is for heating domestic tap water. In addition, the energy content of wastewater is mainly in the form of heat (85%). Hence, there is an obvious interest in recovering a large portion of this heat. However, city-wide scenario analyses that evaluate heat recovery at various locations while considering impacts on wastewater treatment plant (WWTP) performance are currently very limited. This study presents a comprehensive model-based city-wide evaluation considering four different heat recovery locations (appliance, household, precinct and WWTP effluent) for a Swedish city with varying degrees of implementation using an uncertainty-based approach. Results show that heat recovery at the appliance level, with heat exchangers installed at 77% of the showers at domestic households, leads to a mean energy recovery of 127 MWh/day with a 0.25 °C reduction in mean WWTP inlet temperature compared to the default case without heat recovery. The highest mean temperature reduction compared to the default case is 1.5 °C when heat is recovered at the precinct level for 77% of the domestic wastewater flow rate. Finally, the impact on WWTP nitrification capacity is negligible in this case due to its large existing capacity and design.

SF-Net: A Multi-Task Model for Brain Tumor Segmentation in Multimodal MRI via Image Fusion

Automatic segmentation of brain tumor regions from multimodal MRI scans is of great clinical significance. In this letter, we propose a “Segmentation-Fusion” multi-task model named SF-Net for brain tumor segmentation. In comparison to the widely-used multi-task model that adds a variational autoencoder (VAE) decoder to reconstruct the input data, using image fusion as an additional regularization for feature learning helps to achieve more sufficient fusion of multimodal features, which is beneficial to the multimodal image segmentation problem. To further improve the performance of the multi-task model, an uncertainty-based approach that can adaptively adjust the loss weights of different tasks during the training process is introduced for model training. Experimental results on the BraTS 2020 benchmark demonstrate that the proposed method can achieve higher segmentation accuracy than the VAE-based approach. In addition, as the by-product of the multi-task model, the image fusion results obtained are of high quality on the brain tumor regions. The source code of the proposed method is available at https://github.com/yuliu316316/SF-Net.

Assessing the predictive impact of factor fixing with an adaptive uncertainty-based approach

Despite widespread use of factor fixing in environmental modeling, its effect on model predictions has received little attention and is instead commonly presumed to be negligible. We propose a proof-of-concept adaptive method for systematically investigating the impact of factor fixing. The method uses Global Sensitivity Analysis methods to identify groups of sensitive parameters, then quantifies which groups can be safely fixed at nominal values without exceeding a maximum acceptable error, demonstrated using the 21-dimensional Sobol’ G-function. Three error measures are considered for quantities of interest, namely Relative Mean Absolute Error, Pearson Product-Moment Correlation and Relative Variance. Results demonstrate that factor fixing may cause large errors in the model results unexpectedly, when preliminary analysis suggests otherwise, and that the default value selected affects the number of factors to fix. To improve the applicability and methodological development of factor fixing, a new research agenda encompassing five opportunities is discussed for further attention.

Estimation of a Minimum Allowable Structural Strength Based on Uncertainty in Material Test Data.

Three types of uncertainties exist in the estimation of the minimum fracture strength of a full-scale component or structure size. The first, to be called the "model selection uncertainty," is in selecting a statistical distribution that best fits the laboratory test data. The second, to be called the "laboratory-scale strength uncertainty," is in estimating model parameters of a specific distribution from which the minimum failure strength of a material at a certain confidence level is estimated using the laboratory test data. To extrapolate the laboratory-scale strength prediction to that of a full-scale component, a third uncertainty exists that can be called the "full-scale strength uncertainty." In this paper, we develop a three-step approach to estimating the minimum strength of a full-scale component using two metrics: One metric is based on six goodness-of-fit and parameter-estimation-method criteria, and the second metric is based on the uncertainty quantification of the so-called A-basis design allowable (99 % coverage at 95 % level of confidence) of the full-scale component. The three steps of our approach are: (1) Find the "best" model for the sample data from a list of five candidates, namely, normal, two-parameter Weibull, three-parameter Weibull, two-parameter lognormal, and three-parameter lognormal. (2) For each model, estimate (2a) the parameters of that model with uncertainty using the sample data, and (2b) the minimum strength at the laboratory scale at 95 % level of confidence. (3) Introduce the concept of "coverage" and estimate the fullscale allowable minimum strength of the component at 95 % level of confidence for two types of coverages commonly used in the aerospace industry, namely, 99 % (A-basis for critical parts) and 90 % (B-basis for less critical parts). This uncertainty-based approach is novel in all three steps: In step-1 we use a composite goodness-of-fit metric to rank and select the "best" distribution, in step-2 we introduce uncertainty quantification in estimating the parameters of each distribution, and in step-3 we introduce the concept of an uncertainty metric based on the estimates of the upper and lower tolerance limits of the so-called A-basis design allowable minimum strength. To illustrate the applicability of this uncertainty-based approach to a diverse group of data, we present results of our analysis for six sets of laboratory failure strength data from four engineering materials. A discussion of the significance and limitations of this approach and some concluding remarks are included.

Uncertainty analysis of large structures using universal grey number theory

In this paper, a new uncertainty-based approach, termed the universal grey number-based Gaussian elimination method, is presented for the analysis of large structures that require the solution of systems of linear interval algebraic equations. Although exact ranges of the solution can be found using the enumeration method, it is known to be computationally expensive as it requires a large number of analyses. Although interval analysis has been used by some researchers, it is found to lead to wider ranges of the response quantities due to the dependency problem. Hence, modified procedures such as the truncation-based interval analysis have been suggested in the literature to overcome the dependency problem. In fact, no interval analysis-based method is available in the literature for solving large number of interval linear equations accurately. The present method is expected to overcome the limitations associated with the available methods in terms of accuracy and computational effort. To demonstrate the accuracy of the proposed method, the stress analysis of several truss structures under specified interval values of input parameters is considered. It is shown that the proposed method yields accurate results more efficiently with less computational effort compared to the truncation-based interval analysis and enumeration method.

Uncertainty Expression and Propagation in the Risk Assessment of Uncertain Random System

Expressions and propagations of uncertainties have been the core topics in the development of risk science in last decades, whose purpose is to provide decision-makers with clear information about the uncertainty of system's risk. Uncertainty theory is an emerging theory that has great advantages in the expression of epistemic uncertainty, compared to the theories such as subjective probability, evidence theory, and possibility theory. A new framework of uncertainty expression is proposed in this article to properly express different uncertainty sources that exist in the system, where the uncertainty theory is used to express epistemic uncertainties and frequentist probability is used to express aleatory uncertainties. A general propagation method is further developed to address the joint calculations of proposed uncertainty expressions, which can be degenerated into pure probabilistic approach and pure uncertainty-based approach. A case study is conducted to show the effectiveness of proposed method, whose results indicate that the application of proposed method will lead to a result which is easier to be understood than results of theories such as evidence theory and possibility theory, and is more robust than results of Bayesian approach.

Hydraulic fracturing: New uncertainty based modeling approach for process design using Monte Carlo simulation technique.

Hydraulic fracturing is a key method used in completion of shale gas wells as well as in well stimulation. There are a lot of factors affecting the hydraulic fracturing treatment; i.e. formation in-situ stresses, fracturing fluid properties, proppant, pumping rate, reservoir fluid and rock properties…etc. For predictive modeling, these factors are associated with a lot of uncertainties, since most of them are laboratory measured, calculated or subjectively estimated. Moreover, the precise contribution of each factor on the final fracturing result is unknown for each individual case. Therefore, for better treatment performance and in order to find the best range of designing parameters, a hydraulic fracturing predictive model that involves these uncertainties is required specially for newly exploited shale gas reservoir. In this paper a new uncertainty-based approach is described for hydraulic fracturing processes. It is based on assigning probability distribution for some variables and parameters used in the designing process. These probability distributions are used as input data for analytical equations that describe the fracturing processes. Monte Carlo Simulation technique is used to apply uncertainty-based values on the designing analytical formulas. A hypothetical hydraulic fracturing example is used to simulate the effect of different variables and designing parameters on the entire fracturing process. The simulation results are illustrated into probability distribution curves and variance-based sensitivity analysis is performed to assess the contribution and the correlation between different variables and outcomes. Fracture geometry is almost controlled by the injection fluid’s viscosity, in case of constant injection rate; while rock properties have insignificant effect on the fracture width compares to fracturing fluid’s effect. Therefore more emphases shall be directed to rheological modeling of the fracturing fluid. It is found also that fracture height, which is difficult to be estimated, is the most crucial parameter in the calculation of treatment size or the injected fluid’s volume. Proppant porosity, injected fluid viscosity and formation strength are slightly affecting propped fracture width, while proppant final concentration plays the main role of determining the calculated propped fracture width. It is observed from the simulation results that the initial formation permeability will extremely affect the post fracturing skin factor while other formation rock properties have almost no effect on the skin factor. Throughout the implementation of the uncertainty-based modeling approach for hydraulic fracturing process design, it is found that uncertainties in the value of many variables and parameters are slightly affecting the process outcomes. However, injected fluid viscosity, shale formation permeability and proppant final concentration are found to be the most influencing factors in the entire process. Therefore, it is highly recommended to perform in-depth study for these factors prior conducting any designing process of hydraulic fracturing.

Coupled climate–carbon cycle simulation of the Last Glacial Maximum atmospheric CO<sub>2</sub> decrease using a large ensemble of modern plausible parameter sets

Abstract. During the Last Glacial Maximum (LGM), atmospheric CO2 was around 90 ppmv lower than during the pre-industrial period. The reasons for this decrease are most often elucidated through factorial experiments testing the impact of individual mechanisms. Due to uncertainty in our understanding of the real system, however, the different models used to conduct the experiments inevitably take on different parameter values and different structures. In this paper, the objective is therefore to take an uncertainty-based approach to investigating the LGM CO2 drop by simulating it with a large ensemble of parameter sets, designed to allow for a wide range of large-scale feedback response strengths. Our aim is not to definitely explain the causes of the CO2 drop but rather explore the range of possible responses. We find that the LGM CO2 decrease tends to predominantly be associated with decreasing sea surface temperatures (SSTs), increasing sea ice area, a weakening of the Atlantic Meridional Overturning Circulation (AMOC), a strengthening of the Antarctic Bottom Water (AABW) cell in the Atlantic Ocean, a decreasing ocean biological productivity, an increasing CaCO3 weathering flux and an increasing deep-sea CaCO3 burial flux. The majority of our simulations also predict an increase in terrestrial carbon, coupled with a decrease in ocean and increase in lithospheric carbon. We attribute the increase in terrestrial carbon to a slower soil respiration rate, as well as the preservation rather than destruction of carbon by the LGM ice sheets. An initial comparison of these dominant changes with observations and paleoproxies other than carbon isotope and oxygen data (not evaluated directly in this study) suggests broad agreement. However, we advise more detailed comparisons in the future, and also note that, conceptually at least, our results can only be reconciled with carbon isotope and oxygen data if additional processes not included in our model are brought into play.

Redefining the impact assessment of buildings: an uncertainty-based approach to rating codes

ABSTRACTDiscrepancies between predicted and in-use building performance are well documented in impact assessments for buildings, such as rating codes. This is a consequence of uncertainties that undermine predictions, which include procedural errors, as well as users’ behaviour and technological change. Debate on impact assessment for buildings predominantly focuses on operational issues and does not question the deterministic model on which assessments are based as a potential, underlying cause of ineffectiveness. This article builds on a non-deterministic urban planning theory and the principles it outlines, which can help manage uncertain factors over time. A rating code model is proposed that merges its typical steps of assessment (i.e. classification, characterisation and valuation) with those principles, applied within the impact assessment of buildings. These are experimentation (of other criteria than those typically appraised), exploration (the process of identifying the long-term vulnerability of such criteria) and inquiry (iterating and critically evaluating the assessment over time).

An uncertainty-based approach to discrete-time fault estimation observer design for nonuniformly sampled systems

This paper considers the fault estimation problem of nonuniformly sampled system in which sensor sampling is performed at aperiodic interval. After being discretized at sampling instant, the nonuniformly sampled system is modeled as an equivalent polytopic system with norm bounded uncertainties. A discrete-time time-varying fault estimation observer with multiple design freedom is then constructed, and a sufficient condition given in linear matrix inequality (LMI) is provided to obtain the constant filter gain and ensure not only the asymptotic stability of fault estimation error but also the robustness of uncertainties. Compared with the existing observer designed based on continuous-time delay approach, the proposed one has a better estimation accuracy and less conservatism and is easy for digital implementation. A numerical simulation and a quadruple-tank benchmark are used to demonstrate the effectiveness and superiority of the proposed method.