Severe Misclassification Research Articles

Spatio-temporal multi-level attention crop mapping method using time-series SAR imagery

Accurate crop mapping is of great significance for crop yield forecasting, agricultural productivity development and agricultural management. Thanks to its all-time and all-weather capability, integrating multi-temporal synthetic aperture radar (SAR) for crop mapping has become essential and challenging task in remote sensing. In recent years, deep learning (DL) has demonstrated excellent crop mapping accuracy to interpret crop dynamics. However, existing DL-based methods tend to be incapable of capturing spatial and temporal features at different scales simultaneously, and this often leads to severe mis-classification due to the complex and heterogeneous distribution of crops and diverse phenological patterns. In this paper, we propose a novel spatio-temporal multi-level attention method, named as STMA, for crop mapping using time-series SAR imagery in an end-to-end fashion to increase the capability of crop phenology retrieval. Specifically, the multi-level attention mechanism is designed to aggregate multi-scale spatio-temporal representations on crops via cascaded spatio-temporal self-attention (STSA) and multi-scale cross-attention (MCA) modalities. To ensure a fine extraction of multi-granularity features, a learnable spatial attention position encoding is proposed to adaptively generate the position priors to facilitate multi-level attention learning. Experimental results on Brandenburg Sentinel-1 dataset, public PASTIS-R dataset and South Africa dataset demonstrated that STMA can achieve state-of-the-art performance in crop mapping tasks, with the accuracy of 96.54% in the Brandenburg Sentinel-1 dataset, 86.77% in the PASTIS-R dataset and 83.37% in the South Africa dataset, validating its effectiveness and superiority. Further comparison of spatio-temporal generalization capability reflected its excellent performance in spatio-temporal modeling on different crops and scenarios. This research provides a viable and intelligent spatio-temporal framework for large-area crop mapping using time-series SAR imagery in complex agricultural systems. The Brandenburg Sentinel-1 dataset and the STMA code will be publicly available at https://github.com/hanzhu97702/ISPRS_STMA.

Inherit With Distillation and Evolve With Contrast: Exploring Class Incremental Semantic Segmentation Without Exemplar Memory.

As a front-burner problem in incremental learning, class incremental semantic segmentation (CISS) is plagued by catastrophic forgetting and semantic drift. Although recent methods have utilized knowledge distillation to transfer knowledge from the old model, they are still unable to avoid pixel confusion, which results in severe misclassification after incremental steps due to the lack of annotations for past and future classes. Meanwhile data-replay-based approaches suffer from storage burdens and privacy concerns. In this paper, we propose to address CISS without exemplar memory and resolve catastrophic forgetting as well as semantic drift synchronously. We present Inherit with Distillation and Evolve with Contrast (IDEC), which consists of a Dense Knowledge Distillation on all Aspects (DADA) manner and an Asymmetric Region- wise Contrastive Learning (ARCL) module. Driven by the devised dynamic class-specific pseudo-labelling strategy, DADA distils intermediate-layer features and output-logits collaboratively with more emphasis on semantic-invariant knowledge inheritance. ARCL implements region- wise contrastive learning in the latent space to resolve semantic drift among known classes, current classes, and unknown classes. We demonstrate the effectiveness of our method on multiple CISS tasks by state-of-the-art performance, including Pascal VOC 2012, ADE20K and ISPRS datasets. Our method also shows superior anti-forgetting ability, particularly in multi-step CISS tasks.

IPCAPS: an R package for iterative pruning to capture population structure

BackgroundResolving population genetic structure is challenging, especially when dealing with closely related or geographically confined populations. Although Principal Component Analysis (PCA)-based methods and genomic variation with single nucleotide polymorphisms (SNPs) are widely used to describe shared genetic ancestry, improvements can be made especially when fine-scale population structure is the target.ResultsThis work presents an R package called IPCAPS, which uses SNP information for resolving possibly fine-scale population structure. The IPCAPS routines are built on the iterative pruning Principal Component Analysis (ipPCA) framework that systematically assigns individuals to genetically similar subgroups. In each iteration, our tool is able to detect and eliminate outliers, hereby avoiding severe misclassification errors.ConclusionsIPCAPS supports different measurement scales for variables used to identify substructure. Hence, panels of gene expression and methylation data can be accommodated as well. The tool can also be applied in patient sub-phenotyping contexts. IPCAPS is developed in R and is freely available from http://bio3.giga.ulg.ac.be/ipcaps

Application of polarization signature to land cover scattering mechanism analysis and classification using multi-temporal C-band polarimetric RADARSAT-2 imagery

This paper presents a novel method of multi-temporal land cover classification based on the polarization signature (PS). Additionally, the temporal change of the scattering mechanism of crops and other land cover classes is analyzed based on the PS. Firstly, the scattering mechanisms of corn, soybean, wheat, forest and urban over time are analyzed. The correlation coefficient is computed between their PS and the PS of canonical targets to study their relationships with the completely polarized surface, dihedral, dipole and helix scattering targets. The pedestal height (PH) of their PS is also calculated to investigate their un-polarized components such as multiple scattering. Then, a novel multi-temporal supervised binary-tree classification scheme (MTSBTCS) with a criterion that maximizes the power difference of PS (MDPS) is developed (MTSBTCS-MDPS), in which the MDPS is proposed to find the optimum orientation angle, ellipticity angle and image acquisition date to maximize the power difference of two different land cover classes and a binary-tree classifier is constructed for the final multi-temporal classification. Seven fully polarimetric C-band RADARASAT-2 images in Fine Quad (FQ) 21 mode acquired in southwestern Ontario, Canada during the entire growing season from May to September in 2012 are used for the purpose of the scattering mechanism analysis and validation of the MTSBTCS-MDPS. The experiment demonstrates that, the scattering mechanism varies with crop type and also changes greatly for each crop when they are in different growing stages. However, the scattering mechanisms of urban and forest areas remain almost the same. In terms of the classification, comparison of the MDPS with the traditional support vector machine (SVM), random forest (RF), and the Wishart distance (WD), shows that for the single-date image classification, both MDPS and WD achieve much higher accuracy on each date than that of the SVM and RF, but the MDPS consumes more time than that of the WD and its accuracy is also lower. However, when they are applied to the multi-temporal images, both SVM and RF suffer a severe misclassification between forest and urban with their accuracy lower than that of the MTSBTCS-MDPS. In addition, the MTSBTCS-MDPS developed in this paper is much less time-consuming approximately 2.5 times less than that of the MTSBTCS-WD, meanwhile, it also achieves the highest accuracy with its overall accuracy of 87.5% and kappa coefficient of 0.85 among all methods applied.

Generalized partially linear regression with misclassified data and an application to labour market transitions

Large data sets that originate from administrative or operational activity are increasingly used for statistical analysis as they often contain very precise information and a large number of observations. But there is evidence that some variables can be subject to severe misclassification or contain missing values. Given the size of the data, a flexible semiparametric misclassification model would be good choice but their use in practise is scarce. To close this gap a semiparametric model for the probability of observing labour market transitions is estimated using a sample of 20 m observations from Germany. It is shown that estimated marginal effects of a number of covariates are sizeably affected by misclassification and missing values in the analysis data. The proposed generalized partially linear regression extends existing models by allowing a misclassified discrete covariate to be interacted with a nonparametric function of a continuous covariate.

An evidential fusion approach for gender profiling

CCTV (Closed-Circuit TeleVision) systems are broadly deployed in the present world. To ensure in-time reaction for intelligent surveillance, it is a fundamental task for real-world applications to determine the gender of people of interest. However, normal video algorithms for gender profiling (usually face profiling) have three drawbacks. First, the profiling result is always uncertain. Second, the profiling result is not stable. The degree of certainty usually varies over time, sometimes even to the extent that a male is classified as a female, and vice versa. Third, for a robust profiling result in cases that a person’s face is not visible, other features, such as body shape, are required. These algorithms may provide different recognition results - at the very least, they will provide different degrees of certainties. To overcome these problems, in this paper, we introduce an Dempster-Shafer (DS) evidential approach that makes use of profiling results from multiple algorithms over a period of time, in particular, Denoeux’s cautious rule is applied for fusing mass functions through time lines. Experiments show that this approach does provide better results than single profiling results and classic fusion results. Furthermore, it is found that if severe mis-classification has occurred at the beginning of the time line, the combination can yield undesirable results. To remedy this weakness, we further propose three extensions to the evidential approach proposed above incorporating notions of time-window, time-attenuation, and time-discounting, respectively. These extensions also applies Denoeux’s rule along with time lines and take the DS approach as a special case. Experiments show that these three extensions do provide better results than their predecessor when mis-classifications occur.

Short look-back periods in pharmacoepidemiologic studies of new users of antibiotics and asthma medications introduce severe misclassification.

The aim of this study was to quantify the effect of the look back period on the misclassification of new users of antibiotics and asthma medications. We included all children born in Denmark from 1995 through 2006 and all prescriptions of antibiotics and asthma medication from 1995 through 2011. The study period was 2007 through 2011. True new users redeemed their first prescription in the study period whereas prior users redeemed their first prescription before the study period. Look-back periods ranged from 30 days up to 12 years prior to the study period, and we defined new users as those without a prescription in the look-back period. The relative misclassification (RM) was estimated as the number of defined new users divided by the number of true new users. For antibiotics, the RM decreased from 4.75 for look-back periods of 30 days to 2.36 for 2 years and 1.33 for 5 years. For asthma medication, the RM decreased from 2.53 for look-back periods of 30 days to 1.48 for 2 years and 1.20 for 5 years. Older age, male gender, and absence of treatment-related diagnoses were associated with higher RM. Studies applying the new user design are strongly dependent on the available information on prescriptions. For drug classes with intermittent use such as asthma medications, even a 2-year look-back period produced severe misclassification. Excluding children with a prior treatment-related diagnosis can reduce the level of misclassification.

Use of an artificial neural network (ANN) for classifying nursing care needed, using incomplete input data.

In German nursing insurance, the act of classifying the client into four categories of disability is based on legally defined distinct criteria. When classifying deceased persons it is often impossible to collect all the required information. We aimed to determine the ability of an artificial neural network (ANN) to calculate the category of disability, to investigate the response of the ANN to input items of different nature, quantity and data quality, and to estimate the minimum number of training data required. The investigation was conducted as a retrospective observational study. The analysis was based on routine records of 14000 adult clients of the nursing insurance. Several ANNs were trained, varying nature, number and quality of the input items as well as the size of the training data set. Each ANN's classification competence was tested on independent validation data, judging the ANN's conformance to the result of the individual expert assessment, using kappa statistics. Fed with all 30 input items available, the net classified 80% of cases correctly (weighted kappa = 0.78). Using three input items, weighted kappa was 0.63. Severe misclassification (deviation by more than one category in either direction) ranged between 0.2% (all 30 input items) and 3.7% (3/30 items). The less complete the individual input items were, the less accurate was the net's estimate. A 20% rate of missing values was well tolerated. A training set comprising 500 cases was adequate. The input item set inherits redundancy. The ANN's ability to correctly respond to subsets of input items makes it a powerful tool in quality control. In the categorization of deceased persons when only an incomplete input item set is available, the ANN can achieve satisfactory results.

Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods

The validation of dietary assessment instruments is critical in the evaluation of diet as a chronic disease risk factor. The objective was to assess the validity of a self-administered food-frequency questionnaire by comparison with dietary recall, urinary nitrogen excretion, and total energy expenditure data. Over a 1-y period, data from twelve 24-h dietary recalls, a food-frequency questionnaire, and four 24-h urine samples were obtained from 134 study participants of the European Prospective Investigation into Cancer and Nutrition (EPIC) Study in Potsdam, Germany. In a substudy of 28 participants, total energy expenditure from doubly labeled water measurements was assessed. Energy-adjusted, deattenuated correlation coefficients between the questionnaire and the recalls ranged from 0.54 for dietary fiber to 0.86 for alcohol. Cross-classification of quintiles of nutrient intakes from the questionnaire and recalls indicated severe misclassification to be <4%. Reported protein intake correlated with estimated protein excretion (r = 0.46). Energy intake and total energy expenditure were also significantly correlated (r = 0.48); however, all but one subject underreported their energy intake. The magnitude of underreporting varied considerably, by 22% on average, and increased slightly with increasing energy intake. A similar pattern of underreporting was observed when energy intakes from the 24-h dietary recalls were compared with total energy expenditure. These data indicate an acceptable relative validity of the food-frequency questionnaire in this study population. Compared with measurements of total energy expenditure and protein excretion, however, only moderate agreement with both the food-frequency questionnaire and the 24-h dietary recalls was observed.

Color and Firmness Classification of Fresh Market Tomatoes

ABSTRACTA total of 370 tomatoes from two seasons were analyzed using a vision system and three mechanical properties sensors which measured firmness parameters. Multiple linear regression indicated classification based on color and firmness could be applied in practical sorting and improves overall classification. Hue values provided adequate information for classification. The best model (R2= 0.96) based on 13 specific colors yielded severe misclassification of 2.2% for classification into 12 maturity classes and 79% correct classification with all samples classified ± one maturity stage according to USDA standards. A weighted color parameter provided a stable model invariant to changes in lighting conditions and yielded excellent results (R2= 0.89). Quality classification was successfully achieved using a vision and drop impact sensor.