Search

The quantification of emissions of greenhouse gases and air pollutants through the inversion of plumes in satellite images remains a complex problem that current methods can only assess with significant uncertainties. The anticipated launch of the CO2M (Copernicus Anthropogenic Carbon Dioxide Monitoring) satellite constellation in 2026 is expected to provide high-resolution images of CO2 (carbon dioxide) column-averaged mole fractions (XCO2), opening up new possibilities. However, the inversion of future CO2 plumes from CO2M will encounter various obstacles. A challenge is the low CO2 plume signal-to-noise ratio due to the variability in the background and instrumental errors in satellite measurements. Moreover, uncertainties in the transport and dispersion processes further complicate the inversion task. To address these challenges, deep learning techniques, such as neural networks, offer promising solutions for retrieving emissions from plumes in XCO2 images. Deep learning models can be trained to identify emissions from plume dynamics simulated using a transport model. It then becomes possible to extract relevant information from new plumes and predict their emissions. In this paper, we develop a strategy employing convolutional neural networks (CNNs) to estimate the emission fluxes from a plume in a pseudo- XCO2 image. Our dataset used to train and test such methods includes pseudo-images based on simulations of hourly XCO2 , NO2 (nitrogen dioxide), and wind fields near various power plants in eastern Germany, tracing plumes from anthropogenic and biogenic sources. CNN models are trained to predict emissions from three power plants that exhibit diverse characteristics. The power plants used to assess the deep learning model's performance are not used to train the model. We find that the CNN model outperforms state-of-the-art plume inversion approaches, achieving highly accurate results with an absolute error about half of that of the cross-sectional flux method and an absolute relative error of ∼ 20 % when only the XCO2 and wind fields are used as inputs. Furthermore, we show that our estimations are only slightly affected by the absence of NO2 fields or a detection mechanism as additional information. Finally, interpretability techniques applied to our models confirm that the CNN automatically learns to identify the XCO2 plume and to assess emissions from the plume concentrations. These promising results suggest a high potential of CNNs in estimating local CO2 emissions from satellite images. [ABSTRACT FROM AUTHOR]

Local credit cooperatives have long played an important role in local financial services. It has made a significant contribution to agricultural production, farmers' incomes, and the economic development of rural areas. In particular, as a financial instrument serving farmers, microfinance management by local credit cooperatives plays a key role in pursuing profits and fulfilling social responsibility. It was therefore important to obtain effective instruments for combating poverty in rural areas from all walks of society. This paper first outlines the development of microfinance loans in Germany and other countries and describes the current situation and some of the challenges facing local credit cooperatives in financial management. Next, we present the basic concepts of data mining, describe the common methods and key techniques of data mining, analyze and compare the properties of the individual data, and show how the associated mining can actually be performed. Next, we will explain the basic model of microfinance for farmers and some risks in detail and analyze and evaluate the characteristics of these risks in the context of local credit cooperatives. As a result, this paper proposes an improved deep convolutional neural network. The optimized algorithm selects the optimal weight threshold value and different iteration times. The results are fewer errors, the results are closer to the correct data, and the efficiency is better than before. The algorithm is more efficient because errors have been greatly reduced and the time spent on them has been slightly reduced. [ABSTRACT FROM AUTHOR]

During recent years, maintaining scientific and medical databases has turned out to be a major challenge in the medical sector. The medical data regarding the patients comprise diverse diagnostic medical reports and features for such disease need to be recorded carefully for providing high-quality treatments. Therefore, this paper proposes a hybrid deep convolutional neural network (HDCNN)-based COOT approach to predict the risk of diseases. Initially, an improved crossover-based levy flight optimization algorithm (ICLFDO) algorithm is employed in selecting and processing the unstructured textual data. Then later, the HDCNN-COOT approach predicts the diseases more accurately. In addition to this, the classifier predicts whether the patient is at risk of disease in the future or not. The efficiency of the proposed model is evaluated by using a dataset obtained from University Hospital of Ludwig Maximilian University of Munich, Germany. The data collected in the data center include 29,477,035 data items from 36,082 patients. The proposed model is capable of using structured data and uses textual data of the patients. Finally, the higher classification accuracy and improved performance of the classifiers can be viewed using the experimental results conducted on five different datasets. [ABSTRACT FROM AUTHOR]

The quantification of emissions of greenhouse gases and air pollutants through the inversion of plumes in satellite images remains a complex problem that current methods can only assess with significant uncertainties. The anticipated launch of the CO2M satellite constellation in 2026 is expected to provide high-resolution images of CO2 column-averaged mole fractions (XCO2), opening up new possibilities. However, the inversion of future CO2 plumes from CO2M will encounter various obstacles. A challenge is the CO2 plume low signal-to-noise ratio, due to the variability of the background and instrumental errors in satellite measurements. Moreover, uncertainties in the transport and dispersion processes further complicate the inversion task. To address these challenges, deep learning techniques, such as neural networks, offer promising solutions for retrieving emissions from plumes in XCO2 images. Deep learning models can be trained to identify emissions from plume dynamics simulated using a transport model. It then becomes possible to extract relevant information from new plumes and predict their emissions. In this paper, we employ convolutional neural networks (CNN) to estimate the emission fluxes from a plume in a pseudo XCO2 image. Our dataset used to train and test such methods includes pseudo images based on simulations of hourly XCO2, NO2 and wind fields near various power plants in Eastern Germany, tracing plumes from anthropogenic and biogenic sources. CNN models are trained to predict emissions from three power plants that exhibit diverse characteristics. The power plants used to assess the deep learning model's performance are not used to train the model. We find that the CNN model outperforms state of the art plume inversion approaches, achieving highly accurate results with an absolute error about half of that of the cross-sectional flux method. Furthermore, we show that our estimations are only slightly affected by the absence of NO2 fields or a detection mechanism as additional information. Finally, interpretability techniques applied to our models confirm that the CNN automatically learns to identify the XCO2 plume and to assess emissions from the plume concentrations. These promising results suggest a high potential of CNNs in estimating local CO2 emissions from satellite images. [ABSTRACT FROM AUTHOR]

The use of deep learning (DL) with convolutional neural networks (CNN) to monitor surface water can be a valuable supplement to costly and labour-intense standard gauging stations. This paper presents the application of a recent CNN semantic segmentation method (SegNet) to automatically segment river water in imagery acquired by RGB sensors. This approach can be used as a new supporting tool because there are only a few studies using DL techniques to monitor water resources. The study area is a medium-scale river (Wesenitz) located in the East of Germany. The captured images reflect different periods of the day over a period of approximately 50 days, allowing for the analysis of the river in different environmental conditions and situations. In the experiments, we evaluated the input image resolutions of 256 × 256 and 512 × 512 pixels to assess their influence on the performance of river segmentation. The performance of the CNN was measured with the pixel accuracy and IoU metrics revealing an accuracy of 98% and 97%, respectively, for both resolutions, indicating that our approach is efficient to segment water in RGB imagery. [ABSTRACT FROM AUTHOR]

Industry 4.0, a term invented by Wolfgang Wahlster in Germany, is celebrating its 10th anniversary in 2021. Still, the digitalization of the production environment is one of the hottest topics in the computer science departments at universities and companies. Optimization of production processes or redefinition of the production concepts is meaningful in light of the current industrial and research agendas. Both the mentioned optimization and redefinition are considered in numerous subtopics and technologies. One of the most significant topics in these areas is the newest findings and applications of artificial intelligence (AI)—machine learning (ML) and deep convolutional neural networks (DCNNs). The authors invented a method and device that supports the wiring assembly in the control cabinet production process, namely, the Wire Label Reader (WLR) industrial system. The implementation of this device was a big technical challenge. It required very advanced IT technologies, ML, image recognition, and DCNN as well. This paper focuses on an in-depth description of the underlying methodology of this device, its construction, and foremostly, the assembly industrial processes, through which this device is implemented. It was significant for the authors to validate the usability of the device within mentioned production processes and to express both advantages and challenges connected to such assembly process development. The authors noted that in-depth studies connected to the effects of AI applications in the presented area are sparse. Further, the idea of the WLR device is presented while also including results of DCNN training (with recognition results of 99.7% although challenging conditions), the device implementation in the wire assembly production process, and its users' opinions. The authors have analyzed how the WLR affects assembly process time and energy consumption, and accordingly, the advantages and challenges of the device. Among the most impressive results of the WLR implementation in the assembly process one can be mentioned—the device ensures significant process time reduction regardless of the number of characters printed on a wire. [ABSTRACT FROM AUTHOR]

We present an approach for detecting early signs for upcoming forest damages by training a Convolutional Neural Network (CNN) for the pixel-wise prediction of the remaining life-time (RLT) of trees in forests based on Sentinel-2 imagery. We focus on a scenario in which reference data are only available for a related task, namely for a bi-temporal pixel-wise classification of forest degradation. This reference is used to train a CNN for the pixel-wise prediction of forest degradation. In this context, we propose a new sub-sampling-based approach for compensating the effects of a heavy class imbalance in the training data. Using the resulting classification model, we predict semi-labels for images of a Sentinel-2 time series, from which training data for a CNN designed to regress the RLT can be derived after some label cleansing. However, due to data gaps in the time series, e.g. caused by clouds, only intervals can be derived for the target variable to be regressed, and for some training pixels one of the interval limits may even be unknown. Consequently, we propose a new loss function for training a CNN for regressing the RLT that only requires the known interval limits. The method is evaluated on a data set in Germany, covering a time-span of 5 years. We show that the proposed sub-sampling strategy for dealing with strong label imbalance when training the classifier significantly reduces the training time compared to other approaches. We further show that our model predicts the RLT with a maximum error of two months for 80% of the forest pixels that die within one year from the acquisition date of the Sentinel-2 image. [ABSTRACT FROM AUTHOR]

• Soil reflectance composites were generated with Sentinel-2 L2A data. • A multi-channel CNN learning framework for soil organic carbon prediction is proposed. • A dual input deep learning architecture improves the predictions. • The framework produced results that can be interpreted. The Copernicus Sentinel-2 multispectral imagery data may be aggregated to extract large-scale, bare soil, reflectance composites, which enable soil mapping applications. In this paper, this approach was tested in the German federal state of Bavaria, to provide estimations for soil organic carbon (SOC). Different temporal ranges were considered for the generation of the composites, including multi-annual and seasonal ranges. A novel multi-channel convolutional neural network (CNN) is proposed. By leveraging the advantages of deep learning techniques, it utilizes complementary information from different spectral pre-treatment techniques. The SOC predictions indicated little dissimilarity amongst the different composites, with the best performance attained for the six-year composite containing only spring months (RMSE = 12.03 g C · kg−1, R2 = 0.64, RPIQ = 0.89). It has been demonstrated that these outcomes outperform other well-known machine learning techniques. An ablation analysis was accordingly performed to evaluate the interplay of the CNN's different components to disentangle the advantages of each aspect of the proposed framework. Finally, a DUal inPut deep LearnIng architecture, named DUPLICITE, is proposed, which concatenates deep spectral features derived from the CNN mentioned earlier, as well as topographical and environmental covariates through an artificial neural network (ANN) to exploit their complementarity. The proposed approach was demonstrated to provide an improvement in the overall prediction performance (RMSE = 11.60 gC · kg−1, R2 = 0.67, RPIQ = 0.92). [ABSTRACT FROM AUTHOR]

Undocumented buildings are buildings which were built years ago, but were never recorded in official digital cadastral maps. Detection of undocumented buildings is of great importance for urban planning and monitoring. The state of Bavaria, Germany, pursues this task based on high resolution optical data and digital surface models, using semi-automatic detection methods, which suffer from a high false alarm rate. In order to study the influence of sampling strategies on the performance of building detection, we have firstly designed a transferability analysis experiment, which has not been adequately addressed in the current literature. In this experiment, we test whether the trained model from a district contains valuable information for building detection in a different district. It was found that the large-scale building detection results can be considerably improved when training samples are collected from different districts. Based on the building detection results, we propose a novel framework for the detection of undocumented buildings using Convolutional Neural Network (CNN) and official geodata. More specifically, buildings are identified as undocumented, when their pixels in the output of the CNN are predicted as "building", whereas they belong to the "non-building" in the Digital Cadastral Map (DFK). The detected undocumented building pixels are subsequently divided into the class of old or new undocumented building with the aid of a Temporal Digital Surface Model (tDSM) in the stage of decision fusion. By doing so, a seamless map of undocumented buildings is generated for 1/4th of the state of Bavaria, Germany at a spatial resolution of 0.4 m, which has demonstrated the use of CNN for the robust detection of undocumented buildings at large-scale. [ABSTRACT FROM AUTHOR]

The convolutional neural network (CNN) is one of the main algorithms that is applied to deep transfer learning for classifying two essential types of liver lesions; Hemangioma and hepatocellular carcinoma (HCC). Ultrasound images, which are commonly available and have low cost and low risk compared to computerized tomography (CT) scan images, will be used as input for the model. A total of 350 ultrasound images belonging to 59 patients are used. The number of images with HCC is 202 and 148, respectively. These images were collected from ultrasound cases.info (28 Hemangiomas patients and 11 HCC patients), the department of radiology, the University of Washington (7 HCC patients), the Atlas of ultrasound Germany (3 HCC patients), and Radiopedia and others (10 HCC patients). The ultrasound images are divided into 225, 52, and 73 for training, validation, and testing. A data augmentation technique is used to enhance the validation performance. We proposed an approach based on ensembles of the best-selected deep transfer models from the on-the-shelf models: VGG16, VGG19, DenseNet, Inception, InceptionResNet, ResNet, and EfficientNet. After tuning both the feature extraction and the classification layers, the best models are selected. Validation accuracy is used for model tuning and selection. The accuracy, sensitivity, specificity and AUROC are used to evaluate the performance. The experiments are concluded in five stages. The first stage aims to evaluate the base model performance by training the on-the-shelf models. The best accuracy obtained in the first stage is 83.5%. In the second stage, we augmented the data and retrained the on-the-shelf models with the augmented data. The best accuracy we obtained in the second stage was 86.3%. In the third stage, we tuned the feature extraction layers of the on-the-shelf models. The best accuracy obtained in the third stage is 89%. In the fourth stage, we finetuned the classification layer and obtained an accuracy of 93% as the best accuracy. In the fifth stage, we applied the ensemble approach using the best three-performing models and obtained an accuracy, specificity, sensitivity, and AUROC of 94%, 93.7%, 95.1%, and 0.944, respectively. [ABSTRACT FROM AUTHOR]

An ongoing outbreak of monkeypox virus (MPXV) was first reported in the United Kingdom on 6 May 2022. As of 17 November, there had been a total of 80 221 confirmed MPXV cases in over 110 countries. Based on data reported between 6 May and 30 June 2022 in the United Kingdom, Spain, and Germany, we applied a deep learning approach using convolutional neural networks to evaluate the parameters of the 2022 MPXV outbreak. The basic reproduction number (R0) of MPXV was estimated to be 2.32 in the United Kingdom, which indicates the active diffusion of MPXV since the beginning of the outbreak. The data from Spain and Germany produced higher median R0 values of 2.42 and 2.88, respectively. Importantly, the estimated R0 of MPXV in the three countries tends to the previously calculated R0 of smallpox (3.50 to 6.00). Furthermore, the incubation (1/ε) and infectious (1/γ) period was predicted between 9 and 10 days and 4−5 days, respectively. The R0 value derived from MPXV is consistent with the significantly increasing number of cases, indicating the risk of a rapid spread of MPXV worldwide, which would provide important insights for the prevention and control of MPXV epidemic. [ABSTRACT FROM AUTHOR]

For load forecasting, numerous machine learning (ML) approaches have been published. Besides fully connected feedforward neural networks (FFNNs), also called multilayer perceptron, more advanced ML approaches like deep, recurrent or convolutional neural networks or ensemble methods have been applied. However, evaluating the added benefit by novel approaches is difficult. Statistical or rule-based methods constitute a too low benchmark. FFNNs need extensive tuning due to their manifold design choices. To address this issue, a structured, comprehensible five-step FFNN model creation methodology is presented, which constitutes of initial model creation, internal parameter selection, feature engineering, architecture tuning and final model creation. The methodology is then applied to forecast real world heat load data of a hospital in Germany. The forecast constitutes of 192 values (upcoming 48 h in 15 min resolution) and is composed of a multi-model univariate forecasting strategy, with three test models developed at first. As a result, the test models show great similarities which simplifies creation of the remaining models. A performance increase of up to 18% between initial and final models points out the importance of model tuning. As a conclusion, comprehensible model tuning is vital to use FFNN models as benchmark. The effort needed can be reduced by the experience gained through repeated application of the presented methodology. [ABSTRACT FROM AUTHOR]

Colorectal cancer (CRC) is a leading cause of cancer-related deaths worldwide. The best method to prevent CRC is with a colonoscopy. During this procedure, the gastroenterologist searches for polyps. However, there is a potential risk of polyps being missed by the gastroenterologist. Automated detection of polyps helps to assist the gastroenterologist during a colonoscopy. There are already publications examining the problem of polyp detection in the literature. Nevertheless, most of these systems are only used in the research context and are not implemented for clinical application. Therefore, we introduce the first fully open-source automated polyp-detection system scoring best on current benchmark data and implementing it ready for clinical application. To create the polyp-detection system (ENDOMIND-Advanced), we combined our own collected data from different hospitals and practices in Germany with open-source datasets to create a dataset with over 500,000 annotated images. ENDOMIND-Advanced leverages a post-processing technique based on video detection to work in real-time with a stream of images. It is integrated into a prototype ready for application in clinical interventions. We achieve better performance compared to the best system in the literature and score a F1-score of 90.24% on the open-source CVC-VideoClinicDB benchmark. [ABSTRACT FROM AUTHOR]

The massive installation of renewable energy sources together with energy storage in the power grid can lead to fluctuating energy consumption when there is a bi-directional power flow due to the surplus of electricity generation. To ensure the security and reliability of the power grid, high-quality bi-directional power flow prediction is required. However, predicting bi-directional power flow remains a challenge due to the ever-changing characteristics of power flow and the influence of weather on renewable power generation. To overcome these challenges, we present two of the most popular hybrid deep learning (HDL) models based on a combination of a convolutional neural network (CNN) and long-term memory (LSTM) to predict the power flow in the investigated network cluster. In our approach, the models CNN-LSTM and LSTM-CNN were trained with two different datasets in terms of size and included parameters. The aim was to see whether the size of the dataset and the additional weather data can affect the performance of the proposed model to predict power flow. The result shows that both proposed models can achieve a small error under certain conditions. While the size and parameters of the dataset can affect the training time and accuracy of the HDL model. [ABSTRACT FROM AUTHOR]

Storms can cause significant damage to forest areas, affecting biodiversity and infrastructure and leading to economic loss. Thus, rapid detection and mapping of windthrows are crucially important for forest management. Recent advances in computer vision have led to highly-accurate image classification algorithms such as Convolutional Neural Network (CNN) architectures. In this study, we tested and implemented an algorithm based on CNNs in an ArcGIS environment for automatic detection and mapping of damaged areas. The algorithm was trained and tested on a forest area in Bavaria, Germany.. It is a based on a modified U-Net architecture that was optimized for the pixelwise classification of multispectral aerial remote sensing data. The neural network was trained on labeled damaged areas from after-storm aerial orthophotos of a ca. 109 k m 2 forest area with RGB and NIR bands and 0.2-m spatial resolution. Around 10 7 pixels of labeled data were used in the process. Once the network is trained, predictions on further datasets can be computed within seconds, depending on the size of the input raster and the computational power used. The overall accuracy on our test dataset was 92 % . During visual validation, labeling errors were found in the reference data that somewhat biased the results because the algorithm in some instance performed better than the human labeling procedure, while missing areas affected by shadows. Our results are very good in terms of precision, and the methods introduced in this paper have several additional advantages compared to traditional methods: CNNs automatically detect high- and low-level features in the data, leading to high classification accuracies, while only one after-storm image is needed in comparison to two images for approaches based on change detection. Furthermore, flight parameters do not affect the results in the same way as for approaches that require DSMs and DTMs as the classification is only based on the image data themselves, and errors occurring in the computation of DSMs and DTMs do not affect the results with respect to the z component. The integration into the ArcGIS Platform allows a streamlined workflow for forest management, as the results can be accessed by mobile devices in the field to allow for high-accuracy ground-truthing and additional mapping that can be synchronized back into the database. Our results and the provided automatic workflow highlight the potential of deep learning on high-resolution imagery and GIS for fast and efficient post-disaster damage assessment as a first step of disaster management. [ABSTRACT FROM AUTHOR]

Germany is the largest market for natural gas in the European Union, with an annual consumption of approx. 95 billion cubic meters. The German high‐pressure gas pipeline network's length is roughly 40 000 km, which enables highly fluctuating quantities of gas to be transported safely over long distances. Considering that similar amounts of gas are also transshipped through Germany to other EU states, it is clear that Germany's gas transport system is essential to the European energy supply. Since the average velocity of gas in a pipeline is only 25 km/h, an adequate high‐precision, high‐frequency forecasting of supply and demand is crucial for efficient control and operation of such a transmission network. We propose a deep learning model based on spatio‐temporal convolutional neural networks (DLST) to tackle the problem of gas flow forecasting in a complex high‐pressure transmission network. Experiments show that our model effectively captures comprehensive spatio‐temporal correlations through modeling gas networks and consistently outperforms state‐of‐the‐art benchmarks on real‐world data sets by at least 15%. The results demonstrate that the proposed model can deal with complex nonlinear gas network flow forecasting with high accuracy and effectiveness. [ABSTRACT FROM AUTHOR]

In this study we investigate how climate change will directly influence the groundwater resources in Germany during the 21st century. We apply a machine learning groundwater level prediction approach based on convolutional neural networks to 118 sites well distributed over Germany to assess the groundwater level development under different RCP scenarios (2.6, 4.5, 8.5). We consider only direct meteorological inputs, while highly uncertain anthropogenic factors such as groundwater extractions are excluded. While less pronounced and fewer significant trends can be found under RCP2.6 and RCP4.5, we detect significantly declining trends of groundwater levels for most of the sites under RCP8.5, revealing a spatial pattern of stronger decreases, especially in the northern and eastern part of Germany, emphasizing already existing decreasing trends in these regions. We can further show an increased variability and longer periods of low groundwater levels during the annual cycle towards the end of the century. Future groundwater levels in Germany are expected to decrease considerably under the influence of changing climate, exacerbating the trends and patterns already occurring. Simulations also show substantially reduced effects under stringent mitigation scenarios. [ABSTRACT FROM AUTHOR]

Crop yield forecasting depends on many interactive factors, including crop genotype, weather, soil, and management practices. This study analyzes the performance of machine learning and deep learning methods for winter wheat yield prediction using an extensive dataset of weather, soil, and crop phenology variables in 271 counties across Germany from 1999 to 2019. We proposed a Convolutional Neural Network (CNN) model, which uses a 1-dimensional convolution operation to capture the time dependencies of environmental variables. We used eight supervised machine learning models as baselines and evaluated their predictive performance using RMSE, MAE, and correlation coefficient metrics to benchmark the yield prediction results. Our findings suggested that nonlinear models such as the proposed CNN, Deep Neural Network (DNN), and XGBoost were more effective in understanding the relationship between the crop yield and input data compared to the linear models. Our proposed CNN model outperformed all other baseline models used for winter wheat yield prediction (7 to 14% lower RMSE, 3 to 15% lower MAE, and 4 to 50% higher correlation coefficient than the best performing baseline across test data). We aggregated soil moisture and meteorological features at the weekly resolution to address the seasonality of the data. We also moved beyond prediction and interpreted the outputs of our proposed CNN model using SHAP and force plots which provided key insights in explaining the yield prediction results (importance of variables by time). We found DUL, wind speed at week ten, and radiation amount at week seven as the most critical features in winter wheat yield prediction. [ABSTRACT FROM AUTHOR]

Load forecasting is of crucial importance for operations of electric power systems. In recent years, deep learning based methods are emerging for load forecasting because their strong nonlinear approximation capabilities can provide more forecasting precision than conventional statistical methods. However, they usually suffer from some problems, e.g., the gradient vanishment and over-fitting. In order to address these problems, an unshared convolution based deep learning model with densely connected network is proposed. In this model, the backbone is the unshared convolutional neural network and a densely connected structure is adopted, which could alleviate the gradient vanishment. What is more, we use a regularization method named clipped $L_2$ -norm to overcome over-fitting, and design a trend decomposition strategy to address the possible distribution differences between the training and validation data. Finally, we conduct five case studies to verify the outperformance of our proposed deep learning model for deterministic and interval load forecasting. Two high-voltage and an medium-voltage real load datasets from Australia, Germany and America are used for model training and validation, respectively. Results show that the proposed model can achieve higher load forecasting accuracy, compared with other existing methods including the popular conventional methods such as naive forecast and generalized additive model, and deep learning methods, e.g., long short-term memory network, convolutional neural network, fully connected network, etc. [ABSTRACT FROM AUTHOR]

Background: Human evaluation of pathological slides cannot accurately predict lymph node metastasis (LNM), although accurate prediction is essential to determine treatment and follow-up strategies for colon cancer. We aimed to develop accurate histopathological features for LNM in colon cancer. Methods: We developed a deep convolutional neural network model to distinguish the cancer tissue component of colon cancer using data from the tissue bank of the National Center for Tumor Diseases and the pathology archive at the University Medical Center Mannheim, Germany. This model was applied to whole-slide pathological images of colon cancer patients from The Cancer Genome Atlas (TCGA). The predictive value of the peri-tumoral stroma (PTS) score for LNM was assessed. Results: A total of 164 patients with stages I, II, and III colon cancer from TCGA were analyzed. The mean PTS score was 0.380 (± SD = 0.285), and significantly higher PTS scores were observed in patients in the LNM-positive group than those in the LNM-negative group (P < 0.001). In the univariate analyses, the PTS scores for the LNM-positive group were significantly higher than those for the LNM-negative group (P < 0.001). Further, the PTS scores in lymphatic invasion and any one of perineural, lymphatic, or venous invasion were significantly increased in the LNM-positive group (P < 0.001 and P < 0.001). Conclusion: We established the PTS score, a simplified reproducible parameter, for predicting LNM in colon cancer using computer-based analysis that could be used to guide treatment decisions. These findings warrant further confirmation through large-scale prospective clinical trials. [ABSTRACT FROM AUTHOR]

The prediction of near-surface ozone concentrations is important for supporting regulatory procedures for the protection of humans from high exposure to air pollution. In this study, we introduce a data-driven forecasting model named "IntelliO3-ts", which consists of multiple convolutional neural network (CNN) layers, grouped together as inception blocks. The model is trained with measured multi-year ozone and nitrogen oxide concentrations of more than 300 German measurement stations in rural environments and six meteorological variables from the meteorological COSMO reanalysis. This is by far the most extensive dataset used for time series predictions based on neural networks so far. IntelliO3-ts allows the prediction of daily maximum 8 h average (dma8eu) ozone concentrations for a lead time of up to 4 d, and we show that the model outperforms standard reference models like persistence models. Moreover, we demonstrate that IntelliO3-ts outperforms climatological reference models for the first 2 d, while it does not add any genuine value for longer lead times. We attribute this to the limited deterministic information that is contained in the single-station time series training data. We applied a bootstrapping technique to analyse the influence of different input variables and found that the previous-day ozone concentrations are of major importance, followed by 2 m temperature. As we did not use any geographic information to train IntelliO3-ts in its current version and included no relation between stations, the influence of the horizontal wind components on the model performance is minimal. We expect that the inclusion of advection–diffusion terms in the model could improve results in future versions of our model. [ABSTRACT FROM AUTHOR]

Quantitative precipitation estimation with commercial microwave links (CMLs) is a technique developed to supplement weather radar and rain gauge observations. It is exploiting the relation between the attenuation of CML signal levels and the integrated rain rate along a CML path. The opportunistic nature of this method requires a sophisticated data processing using robust methods. In this study we focus on the processing step of rain event detection in the signal level time series of the CMLs, which we treat as a binary classification problem. This processing step is particularly challenging, because even when there is no rain, the signal level can show large fluctuations similar to that during rainy periods. False classifications can have a high impact on falsely estimated rainfall amounts. We analyze the performance of a convolutional neural network (CNN), which is trained to detect rainfall-specific attenuation patterns in CML signal levels, using data from 3904 CMLs in Germany. The CNN consists of a feature extraction and a classification part with, in total, 20 layers of neurons and 1.4×105 trainable parameters. With a structure inspired by the visual cortex of mammals, CNNs use local connections of neurons to recognize patterns independent of their location in the time series. We test the CNN's ability to recognize attenuation patterns from CMLs and time periods outside the training data. Our CNN is trained on 4 months of data from 800 randomly selected CMLs and validated on 2 different months of data, once for all CMLs and once for the 3104 CMLs not included in the training. No CMLs are excluded from the analysis. As a reference data set, we use the gauge-adjusted radar product RADOLAN-RW provided by the German meteorological service (DWD). The model predictions and the reference data are compared on an hourly basis. Model performance is compared to a state-of-the-art reference method, which uses the rolling standard deviation of the CML signal level time series as a detection criteria. Our results show that within the analyzed period of April to September 2018, the CNN generalizes well to the validation CMLs and time periods. A receiver operating characteristic (ROC) analysis shows that the CNN is outperforming the reference method, detecting on average 76 % of all rainy and 97 % of all nonrainy periods. From all periods with a reference rain rate larger than 0.6 mm h -1 , more than 90 % was detected. We also show that the improved event detection leads to a significant reduction of falsely estimated rainfall by up to 51 %. At the same time, the quality of the correctly estimated rainfall is kept at the same level in regards to the Pearson correlation with the radar rainfall. In conclusion, we find that CNNs are a robust and promising tool to detect rainfall-induced attenuation patterns in CML signal levels from a large CML data set covering all of Germany. [ABSTRACT FROM AUTHOR]

In this study, we present RainNet, a deep convolutional neural network for radar-based precipitation nowcasting. Its design was inspired by the U-Net and SegNet families of deep learning models, which were originally designed for binary segmentation tasks. RainNet was trained to predict continuous precipitation intensities at a lead time of 5 min, using several years of quality-controlled weather radar composites provided by the German Weather Service (DWD). That data set covers Germany with a spatial domain of 900km×900 km and has a resolution of 1 km in space and 5 min in time. Independent verification experiments were carried out on 11 summer precipitation events from 2016 to 2017. In order to achieve a lead time of 1 h , a recursive approach was implemented by using RainNet predictions at 5 min lead times as model inputs for longer lead times. In the verification experiments, trivial Eulerian persistence and a conventional model based on optical flow served as benchmarks. The latter is available in the rainymotion library and had previously been shown to outperform DWD's operational nowcasting model for the same set of verification events. RainNet significantly outperforms the benchmark models at all lead times up to 60 min for the routine verification metrics mean absolute error (MAE) and the critical success index (CSI) at intensity thresholds of 0.125, 1, and 5 mmh-1. However, rainymotion turned out to be superior in predicting the exceedance of higher intensity thresholds (here 10 and 15 mmh-1). The limited ability of RainNet to predict heavy rainfall intensities is an undesirable property which we attribute to a high level of spatial smoothing introduced by the model. At a lead time of 5 min , an analysis of power spectral density confirmed a significant loss of spectral power at length scales of 16 km and below. Obviously, RainNet had learned an optimal level of smoothing to produce a nowcast at 5 min lead time. In that sense, the loss of spectral power at small scales is informative, too, as it reflects the limits of predictability as a function of spatial scale. Beyond the lead time of 5 min , however, the increasing level of smoothing is a mere artifact – an analogue to numerical diffusion – that is not a property of RainNet itself but of its recursive application. In the context of early warning, the smoothing is particularly unfavorable since pronounced features of intense precipitation tend to get lost over longer lead times. Hence, we propose several options to address this issue in prospective research, including an adjustment of the loss function for model training, model training for longer lead times, and the prediction of threshold exceedance in terms of a binary segmentation task. Furthermore, we suggest additional input data that could help to better identify situations with imminent precipitation dynamics. The model code, pretrained weights, and training data are provided in open repositories as an input for such future studies. [ABSTRACT FROM AUTHOR]

Deep learning (DL)—in particular convolutional neural networks (CNN)—methods are widely spread in object detection and recognition of remote sensing images. In the domain of DL, there is a need for large numbers of training samples. These samples are mostly generated based on manual identification. Identifying and labelling these objects is very time-consuming. The developed approach proposes a partially automated procedure for the sample creation and avoids manual labelling of rooftop photovoltaic (PV) systems. By combining address data of existing rooftop PV systems from the German Plant Register, the Georeferenced Address Data and the Official House Surroundings Germany, a partially automated generation of training samples is achieved. Using a selection of 100,000 automatically generated samples, a network using a RetinaNet-based architecture combining ResNet101, a feature pyramid network, a classification and a regression network is trained, applied on a large area and post-filtered by intersection with additional automatically identified locations of existing rooftop PV systems. Based on a proof-of-concept application, a second network is trained with the filtered selection of approximately 51,000 training samples. In two independent test applications using high-resolution aerial images of Saarland in Germany, buildings with PV systems are detected with a precision of at least 92.77 and a recall of 84.47. [ABSTRACT FROM AUTHOR]

Energy system modeling is essential in analyzing present and future system configurations motivated by the energy transition. Energy models need various input data sets at different scales, including detailed information about energy generation and transport infrastructure. However, accessing such data sets is not straightforward and often restricted, especially for energy infrastructure data. We present a detection model for the automatic recognition of pipeline pathways using a Convolutional Neural Network (CNN) to address this lack of energy infrastructure data sets. The model was trained with historical low-resolution satellite images of the construction phase of British gas transport pipelines, made with the Landsat 5 Thematic Mapper instrument. The satellite images have been automatically labeled with the help of high-resolution pipeline route data provided by the respective Transmission System Operator (TSO). We have used data augmentation on the training data and trained our model with four different initial learning rates. The models trained with the different learning rates have been validated with 5-fold cross-validation using the Intersection over Union (IoU) metric. We show that our model can reliably identify pipeline pathways despite the comparably low resolution of the used satellite images. Further, we have successfully tested the model's capability in other geographic regions by deploying satellite images of the NEL pipeline in Northern Germany. [ABSTRACT FROM AUTHOR]

The preservation of cultural heritage assets of all kind is an important task for modern civilizations. This also includes tools and instruments that have been used in the previous decades and centuries. Along with the industrial revolution 200 years ago, mechanical and electrical technologies emerged, together with optical instruments. In the meantime, it is not only museums who showcase these developments, but also companies, universities, and private institutions. Gyroscopes are fascinating instruments with a history dating back 200 years. When J.G.F. Bohnenberger presented his machine to his students in 1810 at the University of Tuebingen, Germany, nobody could have foreseen that this fascinating development would be used for complex orientation and positioning. At the University of Stuttgart, Germany, a collection of 160 exhibits is available and in transition towards their sustainable future. Here, the systems are digitized in 2D, 2.5D, and 3D and are made available for a worldwide community using open access platforms. The technologies being used are computed tomography, computer vision, endoscopy, and photogrammetry. We present a novel workflow for combining voxel representations and colored point clouds, to create digital twins of the physical objects with 0.1 mm precision. This has not yet been investigated and is therefore pioneering work. Advantages and disadvantages are discussed and suggested work for the near future is outlined in this new and challenging field of tech heritage digitization. [ABSTRACT FROM AUTHOR]

Undocumented building constructions are buildings or stories that were built years ago, but are missing in the official digital cadastral maps (DFK). The detection of undocumented building constructions is essential to urban planning and monitoring. The state of Bavaria, Germany, uses two semi-automatic detection methods for this task that suffer from a high false alarm rate. To solve this problem, we propose a novel framework to detect undocumented building constructions using a Convolutional Neural Network (CNN) and official geodata, including high resolution optical data and the Normalized Digital Surface Model (nDSM). More specifically, an undocumented building pixel is labeled as "building" by the CNN but does not overlap with a building polygon of the DFK. The class of old or new undocumented building can be further separated when a Temporal Digital Surface Model (tDSM) is introduced in the stage of decision fusion. In a further step, undocumented story construction is detected as the pixels that are "building" in both DFK and predicted results from CNN, but shows a height deviation from the tDSM. By doing so, we have produced a seamless map of undocumented building constructions for one-quarter of the state of Bavaria, Germany at a spatial resolution of 0.4 m, which has proved that our framework is robust to detect undocumented building constructions at large-scale. Considering that the official geodata exploited in this research is advantageous because of its high quality and large coverage, a transferability analysis experiment is also designed in our research to investigate the sampling strategies for building detection at large-scale. Our results indicate that building detection results in unseen areas at large-scale can be improved when training samples are collected from different districts. In an area where training samples are available, local training sampless collection and training can save much time and effort. [ABSTRACT FROM AUTHOR]