Train Multiple Models Research Articles

Development and validation of machine-learning models for monitoring individual behaviors in group-housed broiler chickens

Animals' individual behavior is commonly monitored by live or video observation by a person. This can be labor intensive and inconsistent. An alternative is the use of machine learning-based computer vision systems. The objectives of this study were to 1) develop and optimize machine learning model frameworks for detecting, tracking and classifying individual behaviors of group-housed broiler chickens in continuous video recordings; and 2) use an independent dataset to evaluate the performance of the developed machine leaning model framework for individual poultry behaviors differentiation. Forty-two video recordings from 4 different pens (total video duration = 1,620 min) were used to develop and train multiple models for detecting and tracking individual birds and classifying 4 behaviors: feeding, drinking, active, and inactive. The optimal model framework was used to continuously analyze an external set of 11 videos (duration = 326 min), and the second-by-second behavior of each individual broiler was extracted for the comparison of human observation. After comparison of model performance, the YOLOv5l, out of 5 detection models, was selected for detecting individual broilers in a pen; the osnet_x0_25_msmt17, out of 4 tracking algorithms, was selected to track each detected bird in continuous frames; and the Gradient Boosting Classifier, out of 12 machine learning classifiers, was selected to classify the 4 behaviors. Most of the models were able to keep previously assigned individual identifications of the chickens in limited amounts of time, but lost the identities throughout an examination period (≥4 min). The final framework was able to accurately predict feeding (accuracy = 0.895) and drinking time (accuracy = 0.9) but subpar for active (accuracy = 0.545) and inactive time (accuracy = 0.505). The algorithms employed by the machine learning models were able to accurately detect feeding and drinking behavior but still need to be improved for maintaining individual identities of the chickens and identifying active and inactive behaviors.

An Evidential Mask Transformer for Left Atrium Segmentation

The segmentation of the left atrium (LA) is required to calculate the clinical parameters of the LA, to identify diseases related to its remodeling. Generally, convolutional networks have been used for this task. However, their performance may be limited as a result of the use of local convolution operations for feature extraction. Also, such models usually need extra steps to provide uncertainty maps such as multiple forward passes for Monte Carlo dropouts or training multiple models for ensemble learning. To address these issues, we adapt mask transformers for LA segmentation which effectively use both local and global information, and train them with evidential learning to generate uncertainty maps from the learned Dirichlet distribution, with a single forward pass. We validated our approach on the STACOM 2013 dataset and found that our method can produce better segmentation performance than baseline models, and can identify locations our model’s responses are not trustable.

Performance Evaluation of Predicting IoT Malicious Nodes Using Machine Learning Classification Algorithms

The prediction of malicious nodes in Internet of Things (IoT) networks is crucial for enhancing network security. Malicious nodes can significantly impact network performance across various scenarios. Machine learning (ML) classification algorithms provide binary outcomes ("yes" or "no") to accurately identify these nodes. This study implements various classifier algorithms to address the problem of malicious node classification, using the “SensorNetGuard” dataset. The dataset, comprising 10,000 records with 21 features, was preprocessed and used to train multiple ML models, including Logistic Regression, Decision Tree, Naive Bayes, K-Nearest Neighbors (KNN), and Support Vector Machine (SVM). Performance evaluation of these models followed the ML workflow, utilizing Python libraries such as scikit-learn, Seaborn, Matplotlib, and Pandas. The results indicated that the Naive Bayes classifier outperformed others with an accuracy of 98.1%. This paper demonstrates the effectiveness of ML classifiers in detecting malicious nodes in IoT networks, providing a robust predictive model for real-time application. The “SensorNetGuard” dataset is available on the IEEE data port and Kaggle platform.

Local Evaluation of Large-scale Remote Sensing Machine Learning-generated Building and Road Dataset: The Case of Rwanda

AbstractAccurate and up-to-date building and road data are crucial for informed spatial planning. In developing regions in particular, major challenges arise due to the limited availability of these data, primarily as a result of the inherent inefficiency of traditional field-based surveys and manual data generation methods. Importantly, this limitation has prompted the exploration of alternative solutions, including the use of remote sensing machine learning-generated (RSML) datasets. Within the field of RSML datasets, a plethora of models have been proposed. However, these methods, evaluated in a research setting, may not translate perfectly to massive real-world applications, attributable to potential inaccuracies in unknown geographic spaces. The scepticism surrounding the usefulness of datasets generated by global models, owing to unguaranteed local accuracy, appears to be particularly concerning. As a consequence, rigorous evaluations of these datasets in local scenarios are essential for gaining insights into their usability. To address this concern, this study investigates the local accuracy of large RSML datasets. For this evaluation, we employed a dataset generated using models pre-trained on a variety of samples drawn from across the world and accessible from public repositories of open benchmark datasets. Subsequently, these models were fine-tuned with a limited set of local samples specific to Rwanda. In addition, the evaluation included Microsoft’s and Google’s global datasets. Using ResNet and Mask R‑CNN, we explored the performance variations of different building detection approaches: bottom-up, end-to-end, and their combination. For road extraction, we explored the approach of training multiple models on subsets representing different road types. Our testing dataset was carefully designed to be diverse, incorporating both easy and challenging scenes. It includes areas purposefully chosen for their high level of clutter, making it difficult to detect structures like buildings. This inclusion of complex scenarios alongside simpler ones allows us to thoroughly assess the robustness of DL-based detection models for handling diverse real-world conditions. In addition, buildings were evaluated using a polygon-wise comparison, while roads were assessed using network length-derived metrics.Our results showed a precision (P) of around 75% and a recall (R) of around 60% for the locally fine-tuned building model. This performance was achieved in three out of six testing sites and is considered the lowest limit needed for practical utility of RSML datasets, according to the literature. In contrast, comparable results were obtained in only one out of six sites for the Google and Microsoft datasets. Our locally fine-tuned road model achieved moderate success, meeting the minimum usability threshold in four out of six sites. In contrast, the Microsoft dataset performed well on all sites. In summary, our findings suggest improved performance in road extraction, relative to building extraction tasks. Moreover, we observed that a pipeline relying on a combination of bottom-up and top-down segmentation, while leveraging open global benchmark annotation dataset as well as a small number of samples for fine-tuning, can offer more accurate RSML datasets compared to an open global dataset. Our findings suggest that relying solely on aggregated accuracy metrics can be misleading. According to our evaluation, even city-level derived measures may not capture significant variations in performance within a city, such as lower accuracy in specific neighbourhoods. Overcoming the challenges of complex areas might benefit from exploring alternative approaches, including the integration of LiDAR data, UAV images, aerial images or using other network architectures.

Residual Temporal Convolutional Network With Dual Attention Mechanism for Multilead-Time Interpretable Runoff Forecasting.

As a pivotal subfield within the domain of time series forecasting, runoff forecasting plays a crucial role in water resource management and scheduling. Recent advancements in the application of artificial neural networks (ANNs) and attention mechanisms have markedly enhanced the accuracy of runoff forecasting models. This article introduces an innovative hybrid model, ResTCN-DAM, which synergizes the strengths of deep residual network (ResNet), temporal convolutional networks (TCNs), and dual attention mechanisms (DAMs). The proposed ResTCN-DAM is designed to leverage the unique attributes of these three modules: TCN has outstanding capability to process time series data in parallel. By combining with modified ResNet, multiple TCN layers can be densely stacked to capture more hidden information in the temporal dimension. DAM module adeptly captures the interdependencies within both temporal and feature dimensions, adeptly accentuating relevant time steps/features while diminishing less significant ones with minimal computational cost. Furthermore, the snapshot ensemble method is able to obtain the effect of training multiple models through one single training process, which ensures the accuracy and robustness of the forecasts. The deep integration and collaborative cooperation of these modules comprehensively enhance the model's forecasting capability from various perspectives. Ablation studies conducted validate the efficacy of each module, and through multiple sets of comparative experiments, it is shown that the proposed ResTCN-DAM has exceptional and consistent performance across varying lead times. We also employ visualization techniques to display heatmaps of the model's weights, thereby enhancing the interpretability of the model. When compared with the prevailing neural network-based runoff forecasting models, ResTCN-DAM exhibits state-of-the-art accuracy, temporal robustness, and interpretability, positioning it at the forefront of contemporary research.

SOFB is a comprehensive ensemble deep learning approach for elucidating and characterizing protein-nucleic-acid-binding residues

Proteins and nucleic-acids are essential components of living organisms that interact in critical cellular processes. Accurate prediction of nucleic acid-binding residues in proteins can contribute to a better understanding of protein function. However, the discrepancy between protein sequence information and obtained structural and functional data renders most current computational models ineffective. Therefore, it is vital to design computational models based on protein sequence information to identify nucleic acid binding sites in proteins. Here, we implement an ensemble deep learning model-based nucleic-acid-binding residues on proteins identification method, called SOFB, which characterizes protein sequences by learning the semantics of biological dynamics contexts, and then develop an ensemble deep learning-based sequence network to learn feature representation and classification by explicitly modeling dynamic semantic information. Among them, the language learning model, which is constructed from natural language to biological language, captures the underlying relationships of protein sequences, and the ensemble deep learning-based sequence network consisting of different convolutional layers together with Bi-LSTM refines various features for optimal performance. Meanwhile, to address the imbalanced issue, we adopt ensemble learning to train multiple models and then incorporate them. Our experimental results on several DNA/RNA nucleic-acid-binding residue datasets demonstrate that our proposed model outperforms other state-of-the-art methods. In addition, we conduct an interpretability analysis of the identified nucleic acid binding residue sequences based on the attention weights of the language learning model, revealing novel insights into the dynamic semantic information that supports the identified nucleic acid binding residues. SOFB is available at https://github.com/Encryptional/SOFB and https://figshare.com/articles/online_resource/SOFB_figshare_rar/25499452.

Microgrid F36ault Detection Method Based on Lightweight Gradient Boosting Machine–Neural Network Combined Modeling

The intelligent architecture based on the microgrid (MG) system enhances distributed energy access through an effective line network. However, the increased paths between power sources and loads complicate the system’s topology. This complexity leads to multidirectional line currents, heightening the risk of current loops, imbalances, and potential short-circuit faults. To address these challenges, this study proposes a new approach to accurately locate and identify faults based on MG lines. Initially, characteristic indices such as fault voltage, voltage fundamentals at each MG measurement point, and extracted features like peak voltage values in specific frequency bands, phase-to-phase voltage differences, and the sixth harmonic components are utilized as model inputs. Subsequently, these features are classified using the Lightweight Gradient Boosting Machine (LightGBM), complemented by the bagging (Bootstrap Aggregating) ensemble learning algorithm to consolidate multiple strong LightGBM classifiers in parallel. The output classification results of the integrated model are then fed into a neural network (NN) for further training and learning for fault-type identification and localization. In addition, a Shapley value analysis is introduced to quantify the contribution of each feature and visualize the fault diagnosis decision-making process. A comparative analysis with existing methodologies demonstrates that the LightGBM-NN model not only improves fault detection accuracy but also exhibits greater resilience against noise interference. The introduction of the bagging method, by training multiple base models on the initial classification subset of LightGBM and aggregating their prediction results, can reduce the model variance and prevent overfitting, thus improving the stability and accuracy of fault detection in the combined model and making the interpretation of the Shapley value more stable and reliable. The introduction of the Shapley value analysis helps to quantify the contribution of each feature to improve the transparency and understanding of the combined model’s troubleshooting decision-making process, reduces the model’s subsequent collection of data from different line operations, further optimizes the collection of line feature samples, and ensures the model’s effectiveness and adaptability.

A novel hybrid model based on Empirical Mode Decomposition and Echo State Network for wind power forecasting

Accurate wind power forecasting is essential for (i) the management of wind energy, (ii) increasing the integration of generated power into the electrical grid, and (iii) enhancing maintenance efficiency. This paper proposes a novel hybrid model for short-term (1-hour ahead) wind power forecasting. The model integrates the echo-state network (ESN) architecture with empirical mode decomposition (EMD) to improve forecast accuracy. Unlike existing approaches that use separate models for each decomposed signal after EMD method, proposed model uses a single ESN structure to predict wind power by incorporating all decomposed signals and their past values. The main advantage of this architecture is that it eliminates the need to train multiple models, thereby streamlining the forecasting process. To evaluate the performance of the proposed model, one year of data from the West of Duddon Sands, Barrow, and Horns Power offshore wind farms is utilized. Firstly, the classical ESN model and the EMD-ESN hybrid model are compared for the three datasets. Then, a comprehensive evaluation is performed by comparing the results of the proposed model with commonly used standalone and hybrid forecasting models such as Multi-Layer Perceptron (MLP), Adaptive-Neuro Fuzzy Inference System (ANFIS), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), EMD-LSTM, EMD-BiLSTM, Variational Mode Decomposition based ESN (VMD-ESN), and Wavelet Decomposition based ESN (WD-ESN). The results show that the EMD-ESN hybrid model outperforms implemented models in predicting wind power in all three datasets. Furthermore, the proposed EMD-ESN model for an onshore wind turbine power data in Germany is compared with the SOTA models, such as Transformer, Informer, Autoformer, and Graph Patch Informer (GPI), in the literature. This study highlights the superior predictive capabilities of proposed model, making it a valuable tool for enhancing the accuracy of wind power forecasts, thereby contributing to the reliable integration of wind energy into the power grid.

Abstract 1826: New inhibitors of oncology targets discovered using the affinity-mediated selection of DNA-encoded chemical libraries followed by machine-learning

Abstract X-Chem’s DNA-Encoded Chemical Library platform generates a large amount of binding data for each target it is screened against. Here, we report the application of machine learning to these large datasets with case studies that include the identification of novel inhibitors of two oncology targets. One, ERα, is an estrogen-activated receptor that controls transcription and stimulates breast cell proliferation including in breast cancer. The second, DCAF1, is the substrate-recognition component of an E3 ligase complex that is associated with the proliferation of colon cancer cells. Over 100 billion DNA-Encoded compounds were synthesized and screened for their affinity to each target. The selection output data was then featurized and used to train multiple machine-learned models. Models with demonstrated predictive power were then used to score virtual catalogue compounds which were subsequently acquired and tested. A range of successful data-science strategies will be described including those that lead to double-digit nM inhibitors of each of DCAF1 and ERα. Citation Format: Anthony D. Keefe, Mohamed AbouZleikha, AJ Baghaie, Matthew A. Clark, Marie-Aude Guie, John P. Guilinger, Ryan Walsh, Ying Zhang. New inhibitors of oncology targets discovered using the affinity-mediated selection of DNA-encoded chemical libraries followed by machine-learning [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1826.

Swift Prediction of Battery Performance: Applying Machine Learning Models on Microstructural Electrode Images for Lithium-Ion Batteries

In this study, we investigate the use of artificial neural networks as a potentially efficient method to determine the rate capability of electrodes for lithium-ion batteries with different porosities. The performance of a lithium-ion battery is, to a large extent, determined by the microstructure (i.e., layer thickness and porosity) of its electrodes. Tailoring the microstructure to a specific application is a crucial process in battery development. However, unravelling the complex correlations between microstructure and rate performance using either experiments or simulations is time-consuming and costly. Our approach provides a swift method for predicting the rate capability of battery electrodes by using machine learning on microstructural images of electrode cross-sections. We train multiple models in order to predict the specific capacity based on the batteries’ microstructure and investigate the decisive parts of the microstructure through the use of explainable artificial intelligence (XAI) methods. Our study shows that even comparably small neural network architectures are capable of providing state-of-the-art prediction results. In addition to this, our XAI studies demonstrate that the models are using understandable human features while ignoring present artefacts.

Assimilation of the deep learning-corrected global forecast system fields into the regional model for improving medium-range persistent precipitation forecasts

The error of regional models gradually increases during long-term integration, and it can be reduced with the constraint of large-scale circulation by assimilating global forecast system (GFS) fields to improve the forecast performance. However, with increasing forecast time, the GFS forecast error also significantly increases and varies widely. Directly assimilating GFS fields into the regional model without processing may generate negative impacts, whereas employing postprocessing correction methods can effectively mitigate GFS forecast errors. In this study, the GFS data of European Centre for Medium-Range Weather Forecasts (ECMWF) are first corrected using a convolutional long- and short-term memory network method. The correction model can correct multiple forecasts simultaneously by training only one model, and the correction results are less erroneous than those obtained by training multiple correction models for the different forecast times and pressure layers. The corrected ECMWF forecast fields are then assimilated into Weather Research and Forecasting Model (WRF) using the nudging method. This assimilation method generally improves the forecasts for precipitation above the moderate rainfall level at forecast times of 4–7 days, and reduces the errors of surface wind speed forecasts. This study improves the accuracy of correction models and the forecasts of persistent precipitation weather event with regional model.

A Few-Shot Semi-Supervised Learning Method for Remote Sensing Image Scene Classification

Few-shot scene classification methods aim to obtain classification discriminative ability from few labeled samples and has recently seen substantial advancements. However, the current few-shot learning approaches still suffer from overfitting due to the scarcity of labeled samples. To this end, a few-shot semi-supervised method is proposed to address this issue. Specifically, semi-supervised learning method is used to increase target domain samples; then we train multiple classification models using the augmented samples. Finally, we perform decision fusion of the results obtained from the multiple models to accomplish the image classification task. According to the experiments conducted on two real few-shot remote sensing scene datasets, our proposed method achieves significantly higher accuracy (approximately 1.70% to 4.33%) compared to existing counterparts.

ML‐Augmented Bayesian Optimization of Pain Induced by Microneedles

AbstractMicroneedles (MNs) have emerged as a promising solution for drug delivery and extraction of body fluids. Pain is an important physiological attribute to be examined when designing MNs. There is no known representation of pain with geometric features of a MN despite the focus on experimental work. This study focuses on optimizing MN designs with the aim of minimizing pain through means of machine learning, finite element analysis, and optimization tools. Three distinct approaches are proposed. The first approach involves training multiple regression models on data obtained through finite element analysis in COMSOL. The second approach uses COMSOL's built‐in nonlinear optimization solver. Finally, the third approach utilizes the LiveLink interface between COMSOL and MATLAB, combined with Bayesian optimization. Each approach presents unique strengths and challenges, with the third approach demonstrating significant promise due to its efficiency, practicality, and time‐saving.

Predictive modeling for technology convergence: A patent data-driven approach through technology topic networks

Anticipating technology convergence is crucial for driving innovation and gaining a competitive advantage. Developing a method for forecasting the convergence of previously unrelated technology fields is necessary. Previous studies explored technology convergence using patents but focused on technology classes. However, these prior studies lacked clear insights because they relied on ambiguous technology classes. Therefore, this study seeks to define technology topics with precise technological meanings and develop model predictions using machine learning through technology topic networks to anticipate technology convergence potential. Specifically, it explores the use of text mining techniques to gain insights into technology topics, highlighting best practices and our research findings. This study constructs a network of technology topics derived from patent documents and maps their interactions using a topic network approach. Technology topic networks are constructed through dynamic topic modeling, and features extracted from these networks are used as input for training multiple classification models. By incorporating new link results and leveraging technological influence, the proposed approach can accurately anticipate technology convergence. The results of this research can have significant implications for firms seeking to gain a competitive advantage through innovation and technology convergence and contribute to the establishment of a next-generation R&D planning system.

Predicting Neoadjuvant Treatment Response in Triple-Negative Breast Cancer Using Machine Learning.

Neoadjuvant chemotherapy (NAC) is the standard treatment for early-stage triple negative breast cancer (TNBC). The primary endpoint of NAC is a pathological complete response (pCR). NAC results in pCR in only 30-40% of TNBC patients. Tumor-infiltrating lymphocytes (TILs), Ki67 and phosphohistone H3 (pH3) are a few known biomarkers to predict NAC response. Currently, systematic evaluation of the combined value of these biomarkers in predicting NAC response is lacking. In this study, the predictive value of markers derived from H&E and IHC stained biopsy tissue was comprehensively evaluated using a supervised machine learning (ML)-based approach. Identifying predictive biomarkers could help guide therapeutic decisions by enabling precise stratification of TNBC patients into responders and partial or non-responders. Serial sections from core needle biopsies (n = 76) were stained with H&E and immunohistochemically for the Ki67 and pH3 markers, followed by whole-slide image (WSI) generation. The serial section stains in H&E stain, Ki67 and pH3 markers formed WSI triplets for each patient. The resulting WSI triplets were co-registered with H&E WSIs serving as the reference. Separate mask region-based CNN (MRCNN) models were trained with annotated H&E, Ki67 and pH3 images for detecting tumor cells, stromal and intratumoral TILs (sTILs and tTILs), Ki67+, and pH3+ cells. Top image patches with a high density of cells of interest were identified as hotspots. Best classifiers for NAC response prediction were identified by training multiple ML models and evaluating their performance by accuracy, area under curve, and confusion matrix analyses. Highest prediction accuracy was achieved when hotspot regions were identified by tTIL counts and each hotspot was represented by measures of tTILs, sTILs, tumor cells, Ki67+, and pH3+ features. Regardless of the hotspot selection metric, a complementary use of multiple histological features (tTILs, sTILs) and molecular biomarkers (Ki67 and pH3) resulted in top ranked performance at the patient level. Overall, our results emphasize that prediction models for NAC response should be based on biomarkers in combination rather than in isolation. Our study provides compelling evidence to support the use of ML-based models to predict NAC response in patients with TNBC.

Multi-lead-time short-term runoff forecasting based on Ensemble Attention Temporal Convolutional Network

In the realm of ecological management and human activities within river basins, short-term runoff forecasting plays a pivotal role. Addressing this need, this paper introduces an innovative framework for short-term runoff forecasting: the Ensemble Attention Temporal Convolutional Network (EA-TCN). The cornerstone of this innovation lies in the effective amalgamation of Temporal Convolutional Network (TCN), lightweight attention mechanism, and ensemble learning strategy. This integration synergistically enhances the model’s overall performance in terms of accuracy, efficiency, and robustness. TCN forms the foundation of this framework, where its efficient architecture, characterized by shared parameters and parallel computation, significantly boosts computational efficiency. Its employment of causal and dilated convolutions adeptly captures long-term dependencies within time series inputs. The incorporated lightweight attention mechanism further augments the TCN, enabling EA-TCN to precisely discern complex relationship in temporal data, particularly exhibiting remarkable temporal robustness across various forecasting horizons—a feat challenging for conventional forecasting approaches. Additionally, the integration of the Snapshot ensemble method within the framework allows for simulating the effect of training multiple models through one single training process, thus further elevating the model’s accuracy and robustness. Rigorous ablation and comparative experiments conducted on the US Columbia River dataset substantiate our claims. The results not only validate the individual merits of each component within EA-TCN but also illuminate the significant advantages of their collective application. Our comprehensive assessment unequivocally demonstrates the framework’s exceptional performance in short-term runoff forecasting, positioning it as a state-of-the-art solution in this field. We will further discuss its impact of vocation education in this industry.

The development of an explainable machine learning system for predicting brain atrophies from electronic health records

AbstractBackgroundThe number of elderly living with dementia is projected to dramatically increase, especially in low‐income and middle‐income countries (LMICs). While obtaining brain scans to detect atrophies and significant radiology findings related to dementia might be costly, the use of electronic health record (EHR) for elderly patients could facilitate identifying patients at a high risk based on routinely collected data. This works, evaluates the use of an interpretable machine learning‐based system that predicts brain atrophies and small vessel disease based on EHR data.MethodWe develop a machine learning predictive system to predict atrophies or small vessel disease from retrospective anonymized EHR data recorded for patients above the age of 50 years from a multispecialty outpatient medical center from Ramallah, Palestine. The patient labelling involved using natural language processing to identify cases in unstructured radiology reports for structural brain Magnetic Resonance Imaging (MRI). For input features, we extracted and preprocessed 30 features for the most commonly collected clinical features such as demographics and laboratory results. The dataset was split into training and test sets, where data preprocessing and augmentation, and hyperparameter tuning were applied before training multiple models including Xgboost, Support Vector Machines, Random Forest, and Logistic Regression. To better understand the relative impact each feature had on the model’s prediction, shapely feature importance analysis was conducted for the top performing model.ResultThe top preforming model was Random Forest, with a performance of Area Under the Precision‐Recall Curve (AUPRC) of 0.989, and 0.842 for Area Under Receiving Operating Curve (AUROC). The feature importance analysis revealed that age, sex, Vitamin D, Vitamin B12 and red blood cell distribution width were the features with highest impact on the model’s predictionConclusionPredicting atrophies from EHRs for elderly visiting outpatient clinics could assist in case‐finding for dementia patients and ultimately help identify patients that would benefit from an MRI. In future works, we aim to further investigate building low‐cost diagnostic tools based on a smaller feature set, which could be valuable for facilities from LMICs.

A hybrid approach for the dynamic monitoring and forecasting of NOx emissions in power plants

The concentration of NOx emissions from thermal power plants affects the efficiency of selective catalytic reduction (SCR), and accurate NOx prediction is of great significance for environmental protection. Some literature proves that the prediction accuracy of NOx concentration can be improved by dividing the data into different operating conditions through algorithms or observation and training multiple models separately. However, this finding has some limitations in practice, for the model can’t automatically judge the working condition of the next stage before the data is generated. In order to fill this gap, this study proposed a novel NOx prediction method that first determines the operating conditions based on the power generation plan and then selects the appropriate forecasting model based on the predicted operating conditions. The corresponding model training data for different working conditions were established by referring to the suggestions provided by experts and the screening of mutual information. The system has been deployed at a coal-fired power plant in Yangzhou. Compared with the single model (the model initially deployed in the power plant), the proposed model is improved in four indicators (R2, MAPE, RMSE, and MSE) on the test dataset.

Lightweight Multiattention Recursive Residual CNN-Based In-Loop Filter Driven by Neuron Diversity

13.7pt Many convolutional neural network (CNN)-based in-loop filters have been proposed to improve coding performance. However, considering the single perception scale, high parameter complexity, and the need to train multiple models for various quantization parameters (QPs), the performance and practicability of most existing methods are limited. Inspired by neuron diversity, this paper proposes a lightweight multiattention recursive residual CNN-based in-loop filter that can handle encoded frames with various QP values, frame types (FTs), and temporal layers (TLs) via a single model. First, multiscale features are learned in the neural network and fused with the proposed multidensity block (MDB) and multiscale fusion attention group (MFAG). Second, a recursive structure is adopted to improve the model depth while saving many parameters. The proposed auxiliary parameter fusion attention (APFA) and long-short-term skip connection (LSTSC) models integrate QPs, FTs and TLs into the model while accelerating training. Finally, we propose implementing LMA-RRCNN in parallel with the standard in-loop filter and select the optimal enhanced result in each patch. The experimental results on standard test sequences show that the proposed method achieves on average 13.70% and 11.87% BD rate savings under all-intra and random-access configurations, respectively, outperforming other state-of-the-art approaches.

A Statistical Approach on Estimations of Climate Change Indices by Monthly Instead of Daily Data

Climate change indices (CCI) profoundly contribute to understanding the climate and its change. They are used to present climate change in an easy-to-understand and tangible way, thus, facilitating climate communication. Most of these indices are calculated by daily data but there are also many valuable data sets that consist solely of a monthly temporal frequency. In this paper, we present a method that enables the estimation of specific CCIs from monthly instead of daily data, allowing the expression and examination of data sets consisting solely of monthly parameters through climate change indices. Therefore, we used the ERA5 Land data supplemented by a CMIP6 ssp5-8.5 climate projection to train multiple regression models with different regression functions and selected the best fitting for every grid point. Using a climate projection as a supplement in training the regression functions accounts for climate change and empowers the method’s application in future climate periods. The method includes a simple bias adjustment (delta change). Its output is regridded to ERA5 Land’s 0.1∘ grid, adapting it to the local environment and offering better application in areas with complex terrain using coarse data. Furthermore, the presented method and its regression coefficients can be created from any data set, allowing an even higher spatial resolution than ERA5 Land’s. While the method performs best for the temperature-related indices in warm temperate climates, it works uniformly well for the precipitation-related index maximum consecutive dry days on a global scale.