Search

The imperative to decarbonize the energy sector has prompted substantial advancements in clean electricity generation, with hydrogen emerging as a promising low-carbon energy carrier. While hydrogen synthesis from renewable sources is crucial, challenges persist, necessitating innovative approaches for efficient and sustainable production. This study leverages diverse artificial neural network (ANN) models to assess and predict system efficiency based on key operational variables in membrane reactor systems. The multilayered perceptron (MLP) and radial basis function (RBF) methodologies are employed, with the MLP models optimized across twelve training algorithms and eight activation functions, exploring up to three hidden layers with variable neuron counts. The MLP model, utilizing the Levenberg-Marquard training algorithm and Tangent-Sigmoid activation function, achieved a high correlation coefficient (R2) of 0.9975 for training and 0.9962 for testing, and a mean squared error (MSE) of 0.00425 for training and 0.23951 for testing, indicating precise and accurate efficiency predictions. The Log-Sigmoid activation function also performed well, with R² values of 0.9971 (training) and 0.9961 (testing), and MSE values of 0.004086 (training) and 0.17694 (testing). Optimization of the RBF network identified the best performance with a spread parameter of 1 and 35 neurons, although the MLP model demonstrated superior accuracy and reduced computational time. Statistical analysis, encompassing correlation coefficient, mean squared error, Root Mean Squared error, absolute average deviation, absolute average relative deviation, and runtime, confirms the network's consistent and accurate estimation of system efficiency across various input variables. The study highlights that applying tansig and logsig activation functions, configured with neuron counts of 20, 17, 6 and 23, 20, 2 at the first, second and third hidden layers, respectively, offers enhanced accuracy and reliability. The MLP model's high performance underscores its potential to identify optimal conditions for H2 generation based on system efficiency, thereby advancing membrane reactor technology for hydrogen production. [ABSTRACT FROM AUTHOR]

Water quality monitoring of rivers is necessary in order to properly manage their basins so that steps can be taken to control the amount of pollutants and bring them to the allowable level. The ARIMA (autoregressive integrated moving average) model does not consider nonlinear patterns in modeling water quality components. Also, in modeling using the MLP (Multilayer Perceptrons) model, both linear and nonlinear pattern are not controlled equally. Therefore, in the present study, linear time series models (ARIMA), MLP model, and a hybrid model of MLP and ARIMA optimized by a Grasshopper optimization algorithm are used to predict water quality components in the statistical period of 2011–2019. In the proposed hybrid method, the ability of the ARIMA and the MLP model are exploited. Observational water quality data for forecasting time series in the hybrid method include dissolved oxygen, water temperature, and boron over 108 months. Since, the hybrid model is capable of realizing the nonlinear essence of complicated time series, it makes more reliable forecasts. In the hybrid model, the correlation coefficients between the observational data and the predicted values are 0.9 for dissolved oxygen, 0.91 for water temperature, and 0.91 for boron. To compare the three ARIMA, MLP, and hybrid models, the accuracy indices of each model are calculated. The results show that the hybrid model's higher accuracy compared with the other two models. [ABSTRACT FROM AUTHOR]

Background & purpose: Tremor is a common movement disorder diagnosed employing electrophysiological methods. Today, machine learning (ML) algorithms can efficiently analyze image-based data. Thus, we subjected the dynamics of essential tremor (ET) to video-based analysis. Methods: We enrolled 59 ET patients and 48 age-matched normal controls. The Clinical Rating Scale for Tremor was used to score tremors. All subjects used a smartphone to record an image designed especially for this study while both stationary and in motion. The trajectories were divided into lower bandpass-filtered and bandpass-filtered (BPF) groups based on the frequency. We extracted seven trajectory features, including the angle, velocity, homogeneity, pitch, power, entropy, and cosine. We used Student's t-test to compare the features of the ET patients and normal controls. A Random Forest model was employed to rank feature importance. Five ML models (random forest, k-nearest neighbors, support vector machine, decision tree, and multi-layer perceptron) were applied to estimate diagnostic accuracy. Results: Significant differences in most of the features of the BPF signals were evident between the two groups. The velocity and homogeneity of the BPF trajectory were highest in the stationary and motion phases, respectively. The highest accuracy levels in the stationary, motion, and combined phases for predicting ET were 0.901, 0.757, and 0.892, respectively. Conclusions: Features of ET tremor were evident in image-based data, enabling analysis of the tremor dynamics. ML analyses distinguished ET subjects from normal controls; however, more research is needed. [ABSTRACT FROM AUTHOR]

Volume visualization plays a crucial role in both academia and industry, as volumetric data is extensively utilized in fields such as medicine, geosciences, and engineering. Addressing the complexities of volume rendering, neural rendering has emerged as a potential solution, facilitating the production of high-quality volume rendered images. In this paper, we propose TwineNet, a neural network architecture specifically designed for volume rendering. TwineNet combines features extracted from volume data, transfer functions, and viewpoints by utilizing twining skip connections across multiple feature layers. Building upon the TwineNet architecture, we introduce two neural networks, VolTFNet and PosTFNet, which leverage convolutional encoder–decoders and multilayer perceptrons to synthesize volume rendered images with novel transfer functions and viewpoints. Our experimental findings demonstrate the superiority of our models compared to state-of-the-art approaches in generating high-quality volume rendered images with novel transfer functions and viewpoints. This research contributes to advancing the field of volume rendering and showcases the potential of neural rendering techniques in scientific visualization. [ABSTRACT FROM AUTHOR]

Most existing knowledge graphs (KGs) suffer from incompleteness, which will be detrimental to a variety of downstream applications. Link prediction is the task of predicting missing links in the KGs and can effectively address the issue of incompleteness by knowledge graph embedding (KGE). ConvE, a relatively popular KGE model based on convolutional neural networks, has shown superiority in link prediction. Some subsequent extension models of ConvE achieve state-of-the-art performance by increasing complexity and training time, which result in a high risk of overfitting and a limited performance due to the large number of parameters concentrated in the fully connected projection layer. To address these challenges, we for the first time innovatively introduce and extend a recently simple network architecture gMLP (based on multi-layer perceptrons MLPs with gating) in vision applications for link prediction. We propose a simple and efficient model called gMLP-KGE, which consists of an embedding layer, an input layer, an extended gMLP layer, and an output layer. Extensive experiments show that the number of parameters of gMLP-KGE is close to that of ConvE and less than other extension models, while gMLP-KGE consistently performs well on seven datasets of different scales under most evaluation metrics. [ABSTRACT FROM AUTHOR]

The bitterling fish is a prime example of intelligent behavior in nature for survival. The bitterling fish uses the oyster spawning strategy as their babysitter. The female bitterling fish looks for a male fish stronger than other fish to find the right pair. In order to solve optimization issues, the Bitterling Fish Optimization (BFO) algorithm is modeled in this manuscript based on the mating behavior of these fish. The bitterling fish optimization algorithm is more accurate than the gray wolf optimization algorithm, whale optimization algorithm, butterfly optimization algorithm, Harris Hawks optimization algorithm, and black widow optimization algorithm, according to experiments and implementations on various benchmark functions. Data mining and machine learning are two areas where meta-heuristic techniques are frequently used. In trials, the MLP artificial neural network and a binary version of the BFO algorithm are used to lower the detection error for intrusion traffic. The proposed method's accuracy, precision, and sensitivity index for detecting network intrusion are 99.14%, 98.87%, and 98.85%, respectively, according to experiments on the NSL KDD data set. Compared to machine learning approaches like NNIA, DT, RF, XGBoot, and CNN, the proposed method is more accurate at detecting intrusion. The BFO algorithm is used for feature selection in the UNSW-NB15 dataset, and the tests showed that the accuracy of the proposed method is 96.72% in this dataset. The proposed method of the BFO algorithm is also used to improve Kmeans clustering, and the tests performed on the dataset of covid 19, diabetes, and kidney disease show that the proposed method performs better than iECA*, ECA*, GENCLUST + + (G + +) methods. Deep has KNN, LVQ, SVM, ANN, and KNN. [ABSTRACT FROM AUTHOR]

The declining fertility rate and increasing marriage age among girls pose challenges for policymakers, leading to issues such as population decline, higher social and economic costs, and reduced labor productivity. Using machine learning (ML) techniques to predict the desire to have children can offer a promising solution to address these challenges. Therefore, this study aimed to predict the childbearing tendency in women on the verge of marriage using ML techniques. Data from 252 participants (203 expressing a "desire to have children" and 49 indicating "reluctance to have children") in Abadan, and Khorramshahr cities (Khuzestan Province, Iran) was analyzed. Seven ML algorithms, including multilayer perceptron (MLP), support vector machine (SVM), logistic regression (LR), random forest (RF), J48 decision tree, Naive Bayes (NB), and K-nearest neighbors (KNN), were employed. The performance of these algorithms was assessed using metrics derived from the confusion matrix. The RF algorithm showed superior performance, with the highest sensitivity (99.5%), specificity (95.6%), and receiver operating characteristic curve (90.1%) values. Meanwhile, MLP emerged as the top-performing algorithm, showcasing the best overall performance in accuracy (77.75%) and precision (81.8%) compared to other algorithms. Factors such as age of marriage, place of residence, and strength of the family center with the birth of a child were the most effective predictors of a woman's desire to have children. Conversely, the number of daughters, the wife's ethnicity, and the spouse's ownership of assets such as cars and houses were among the least important factors in predicting this desire. ML algorithms exhibit excellent predictive capabilities for childbearing tendencies in women on the verge of marriage, highlighting their remarkable effectiveness. This capacity to offer accurate prognoses holds significant promise for advancing research in this field. [ABSTRACT FROM AUTHOR]

This study aimed to establish a deep learning-based predictive model for the prognosis of idiopathic sudden sensorineural hearing loss (SSNHL). Data from 1108 patients with SSNHL between January 2015 and May 2023 were retrospectively analyzed. Patients underwent standardized treatment protocols including high-dose steroid therapy and hearing outcomes were assessed after three months using Siegel's criteria and the American Academy of Otolaryngology-Head and Neck Surgery (AAO-HNS) classification. For predicting patient recovery, a two-layered classification process was implemented. Initially, a set of 22 Multilayer Perceptrons (MLP) networks was employed to categorize the patients. The outcomes from this initial categorization were subsequently relayed to a second-layer meta-classifier for final prognosis determination. The validity of this methodology was ascertained through a K-fold cross-validation procedure executed with 10 distinct splits. The prediction model for complete recovery, based on Siegel's criteria, demonstrated an accuracy of 0.892 and area under the curve (AUC) of 0.922. For the class A prediction, according to AAO-HNS classification, the model showed an accuracy of 0.847 and AUC of 0.918. These results suggest that the model may have the potential to contribute to the establishment of tailored patient management strategies by predicting hearing recovery in patients with SSNHL. [ABSTRACT FROM AUTHOR]

The achievement of consistent molding quality, which is critical in injection molding, is heavily reliant on the reasonable control of processing materials, molds, machines, process parameters, and environmental conditions. Notably, new molds usually require a trial molding process before being delivered to relevant machines for online production. However, performance differences between machines make it challenging to maintain consistent molding quality, and suitable adjustments must be made to machine parameters to compensate for these differences. Therefore, cross-machine product quality prediction is critical for accurately forecasting product quality across different machines in a manufacturing process and thus for ensuring consistent quality, few defects, and optimized production. To avoid the considerable time and high cost required for quality inspection and to improve production efficiency, this study developed a multilayer perceptron (MLP) model combined with quality indices to predict molding quality. This paper describes how the developed model predicts product quality for the same mold in different machines. The procedure of the proposed MLP model involves four steps. First, data are prepared, features are extracted (extraction of quality indices), and the model is trained on an actual injection molding machine (machine A). Second, the developed MLP model establishes the relationships between the process parameters, quality indices, and product quality for machine A. Third, Moldex3D Studio, which is a software program for simulating injection molding, is employed to generate production data for a virtual injection molding machine (machine B). Finally, a transfer model is used to fit the quality indices of machines A and B so that the MLP model can directly predict the product quality (in terms of weight and geometric dimensions) for machine B on the basis of the quality indices generated using the process parameters of machine B. Experimental results indicate that the developed MLP model can accurately predict the weight and dimensions of products manufactured using different injection molding machines. In particular, the average error in predicting the product quality for machine B was found to be smaller than 0.5%, which indicates the feasibility of the developed model for cross-machine product quality prediction. [ABSTRACT FROM AUTHOR]

A four-step dividing algorithm of sampling points for artificial neural networks is presented to select a representative training subset for modelling the spatial distribution. The chromium and manganese contents in the topsoil in Tarko-Sale and Noyabrsk cities (Russian subarctic zone) were used as raw data. The spatial distributions of the content of elements in the topsoil layer were performed using a multilayer perceptron (MLP) with sigmoid and hyperbolic tangential activation functions. The root means squared error (RMSE) was calculated for each element content and area. The MLP with hyperbolic tangential activation function showed better accuracy for both subarctic cities and model areas. For Noyabrsk, the model with a hyperbolic tangent was about 10% better. For Tarko-Sale, the improvement of RMSE was around 200%. We have identified three classes of points: «elite», «middle», and «useless». Given this information when dividing the raw set, the accuracy of the models will increase. [ABSTRACT FROM AUTHOR]

Unsupervised cross-modal hashing (UCMH) has been commonly explored to support large-scale cross-modal retrieval of unlabeled data. Despite promising progress, most existing approaches are developed on convolutional neural network and multilayer perceptron architectures, sacrificing the quality of hash codes due to limited capacity for excavating multi-modal semantics. To pursue better content understanding, we break this convention for UCMH and delve into a transformer-based paradigm. Unlike naïve adaptations via backbone substitution that overlook the heterogeneous semantics from transformers, we propose a multi-granularity learning framework called hugging to bridge the modality gap. Specifically, we first construct a fine-grained semantic space composed of a series of aggregated local embeddings that capture implicit attribute-level semantics. In the hash learning stage, we innovatively incorporate fine-grained alignment with these local embeddings to enhance global hash code alignment. Notably, this fine-grained alignment only facilitates robust cross-modal learning without complicating global hash code generation at test time, thus fully maintaining the high efficiency of hash-based retrieval. To make the most of fine-grained information, we further propose a differentiable optimized quantization algorithm and extend our framework to hugging + . This variant neatly integrates quantization learning into the fine-grained alignment during training, producing quantization codes of local embeddings as a gift at test time, which can augment the retrieval performance through an efficient reranking stage. We instantiate simple baselines with contrastive learning objectives for hugging and hugging + , namely HuggingHash and HuggingHash + . Extensive experiments on 4 text-image retrieval and 2 text-video retrieval benchmark datasets show the competitive performance of HuggingHash and HuggingHash + against state-of-the-art baselines. More encouragingly, we also validate that hugging and hugging + are flexible and effective across various baselines, suggesting their universal applicability in the realm of UCMH. [ABSTRACT FROM AUTHOR]

Alzheimer's detection is a challenging task for physicians. There are subtle differences in the bio-marker characteristics of Alzheimers and mild cognitive impairment patients which is very difficult to detect by a physician. Machine learning approaches are widely used for predicting a patient as having either Alzheimer's or mild cognitive impairment. For developing models that distinguish between Alzheimer's and mild cognitive impairment patients, the researchers used a dynamic ensemble of classifier selection algorithms. These algorithms perform voting on ensemble classifiers without considering preferential choices of the Alzheimer's and mild cognitive impairment categories. Thus, this paper applies a modified Borda count voting weightage method instead of the majority voting and Borda voting for classifying Alzheimer's, healthy control, and mild cognitive impairment patients classification on dynamic ensemble of classifier selection algorithms. Six dynamic ensemble of classifier selection algorithms are used in the study. Ten pools of classifiers including random forest, bagged decision tree, extra trees, Adaboost, rotation forest, decision tree, bagged support vector machine, bagged multilayer perceptrons, majority voting ensemble, and stacking classifier are used as classifier input for the dynamic ensemble of classifiers. The results suggest that the application of the proposed method can improve the classification performance for Alzheimer's, mild cognitive impairment, and healthy patients when compared to the traditional voting methods after applying most of the dynamic ensemble of classifier selection algorithms used in the study. The application of a modified Borda count voting method on the dynamic ensemble of classifiers resulted in an increase of balanced classification accuracy ranging from 1 to 9%. The highest balanced classification accuracy of 86% is reported when random forest is applied to meta-learning for dynamic ensemble selection algorithms with the proposed voting method. It is also noted that there is an maximum increase in balanced classification accuracy of 9% is observed when applying rotation forest on K-nearest output profiles classifier using the proposed modified Borda count voting method. Thus, the increase in the balanced classification accuracy after applying the proposed modified Borda count voting method can make a positive impact on high-stakes healthcare applications like Alzheimer's detection. [ABSTRACT FROM AUTHOR]

Enhancement in both energy efficiency and spectral efficiency in cellular networks is made possible by means of an advanced technique called device-to device (D2D) communication. The enhancement of these efficiencies is done by utilizing the cellular user (CU) resources once again to make communication with nearby cellular devices in an effective manner by spectral means. As the D2D communication technology is capable of providing a direct communication link with nearby devices effectively with enhanced spectral efficacy, this approach is considered as an ideal solution for the futuristic cellular communication network. A flexible and reliable relay-assisted communication by means of D2D technology is required that acts as an intermediate relay when the attenuation between the channels across the D2D devices becomes high. The throughput of the system is increased by utilizing D2D communication technology as it uses direct data transmission within cellular devices. When the cellular user is far apart from one another, the data loss in the D2D communication system is minimized with the utilization of the relay. When the channels are good, then the relay nodes (RNs) serve the cellular user. However, it is noted that the D2D systems are affected by issues such as higher consumption of energy and spectral sharing. Also, the sum rate gets degraded as a result of mutual interference between resource-sharing cellular devices in the relay-assisted D2D communication system. The transmission of the data in a traditional relay-assisted D2D communication system is carried out between the D2D receiver (DR) and D2D transmitter (DT) only with the utilization of its own energy by the RN. The issues in relay selection and resource allocation in the conventional joint resource allocation schemes are tackled by executing a scheme for performing the task of optimal relay selection and joint resource allocation. The enhancement of the overall sum rate in the D2D communication is the main motive behind the implemented scheme. This goal is attained along with the minimization of the link rates in the cellular and D2D networks. The ideal selection of the relay and the execution of the joint resource allocation are done with the utilization of a new optimization scheme called the hybrid flow direction with the chameleon swarm algorithm (HFDCSA), in which the flow direction algorithm (FDA) is fused along with chameleon swarm algorithm. This optimal selection of the relays is assisted by considering constraints like the network's sum rate and energy efficiency in the network to achieve high performance. The data obtained from distinct sources are given to the adaptive multi-layer perceptron (AMLP) in which optimal resource allocation and the relay selection are performed with the help of the suggested HFDCSA. The parameters in the MLP are tuned by the same HFDCSA. Finally, the performance validation is conducted in the stage to verify the working of the suggested approach. [ABSTRACT FROM AUTHOR]

This study presents a novel method leveraging surface wave-assisted photonic spin Hall effect (PSHE) to construct physical unclonable functions (PUFs). PUFs exploit inherent physical variations to generate unique Challenge–Response pairs, which are critical for hardware security and arise from manufacturing discrepancies, device characteristics, or timing deviations. We explore PSHE generation-based PUF design, expanding existing design possibilities. With recent applications in precise sensing and computing, PSHE offers promising performance metrics for our proposed PUFs, including an inter-Hamming distance of 47.50% , an average proportion of unique responses of 62.5% , and a Pearson correlation coefficient of − 0.198. The PUF token demonstrates robustness to simulated noise. Additionally, we evaluate security using a machine learning-based attack model, employing a multi-layer perceptron (MLP) regression model with a randomized search method. The average accuracy of successful attack prediction is 9.70% for the selected dataset. Our novel PUF token exhibits high non-linearity due to the PSHE effect, resilience to MLP-based attacks, and sensitivity to process variation. [ABSTRACT FROM AUTHOR]

The increasing reliance on internet-based services has heightened the vulnerability of network infrastructure to cyberattacks, particularly distributed denial of service (DDoS) attacks. These attacks can cause severe disruptions and significant financial losses. Early detection of malicious traffic is crucial in effectively combating such threats. This paper presents an innovative approach called the Encoder-Stacked deep neural networks (E-SDNN) model, which leverages Stacked/bagged multi-layer perceptrons (MLP) for accurate DDoS attack detection. The proposed method employs an encoder to select pertinent features from a preprocessed dataset, enabling precise attack detection. Extensive experiments were conducted on benchmark cybersecurity datasets, namely CICDS2017 and CICDDoS2019, encompassing various DDoS attack scenarios. The experimental results demonstrate the superiority of the E-SDNN model compared to state-of-the-art methods. The proposed E-SDNN model achieved an impressive overall accuracy rate of 99.94% and 98.86% for CICDDS2017 and CICDDoS2019, respectively. [ABSTRACT FROM AUTHOR]

The classification of land use and land cover (LULC) from remotely sensed imagery in semi-arid Mediterranean areas is a challenging task due to the fragmentation of the landscape and the diversity of spatial patterns. Recently, the use of deep learning (DL) for image analysis has increased compared to commonly used machine learning (ML) methods. This paper compares the performance of four algorithms, Random Forest (RF), Support Vector Machine (SVM), Multilayer Perceptron (MLP) and Convolutional Network (CNN), using multi-source data, applying an exhaustive optimisation process of the hyperparameters. The usual approach in the optimisation process of a LULC classification model is to keep the best model in terms of accuracy without analysing the rest of the results. In this study, we have analysed such results, discovering noteworthy patterns in a space defined by the mean and standard deviation of the validation accuracy estimated in a 10-fold cross validation (CV). The point distributions in such a space do not appear to be completely random, but show clusters of points that facilitate the discovery of hyperparameter values that tend to increase the mean accuracy and decrease its standard deviation. RF is not the most accurate model, but it is the less sensitive to changes in hyperparameters. Neural Networks, tend to increase commission and omission errors of the less represented classes because their optimisation lead the model to learn better the most frequent classes. On the other hand, RF and MLP prediction layers are the most accurate from a general qualitative point of view. [ABSTRACT FROM AUTHOR]

Traditional machine learning classifiers can automatically evaluate human behaviour and emotion recognition tasks. However, prior research work does not secure users' privacy and personal information because they need complete access to sensitive physiological data. The recently introduced Federated Learning (FL) paradigm can address this problem. FL allows the local model updates to be sent to a central server, combining them to create a global model. It does not allow the global model to access the raw data used to train it. Motivated by the core concept of FL, this paper proposes a novel FL-based Multi-modal Emotion Recognition System (F-MERS) framework combining EEG, GSR, ECG, and RESP physiological sensors data. It uses Multi-layer Perceptron (MLP) as a base model for classifying complex emotions in three dimensions: Valence, Arousal, and Dominance (VAD). The work validates the F-MERS framework with three emotion benchmark datasets, DEAP, AMIGOS, and DREAMER, achieving accuracies of 87.90%, 89.02%, and 79.02%, respectively. It is the first FL-enabled framework for recognizing complex emotions in three dimensions (VAD) with multi-modal physiological sensors. The proposed study assesses the F-MERS framework in two scenarios: (1). Subject dependent and (2). Subject independent, making the framework more generalized and robust. The experimental outcomes indicate that the F-MERS framework is scalable, efficient in communication, and offers privacy preservation over the baseline Non-FL MLP model. [ABSTRACT FROM AUTHOR]

The present investigation utilized a machine learning structure to ascertain the pressure drop in vertically positioned, corrugated copper tubes during the evaporation process of R134a. The evaporator was a counter-flow heat exchanger, in which R134a flowed in the inner corrugated tube and hot water flowed in the smooth annulus. Different evaporation mass fluxes (195–406 kg m-2 s-1) and heat fluxes (10.16–66.61 kW m-2) were used with artificial neural networks at different corrugation depths. A multilayer perceptron artificial neural network model with 13 neurons in the hidden layer was proposed. Tan-Sig and Purelin transfer functions were used in the network model developed with the Levenberg–Marquardt training algorithm. The dataset, which consisted of 252 data points, related to the evaporation process, was divided into training (70%), validation (15%), and testing (15%) groups in an arbitrary manner. The artificial neural network model has been demonstrated to effectively forecast the pressure drop that occurs during evaporation. The mean squared error was computed for the ΔP values observed during the evaporation processes, yielding a value of 1.96E-03. The artificial neural network exhibited a high correlation coefficient value of 0.94479. The estimation fluctuations exhibited a range of ± 10%, whereas the experimental and anticipated ΔP data demonstrated a divergence of ± 10.3%. [ABSTRACT FROM AUTHOR]

In real-world unbounded outdoor scenes with cars, there are various specular reflections caused by the surrounding environment appearing on the reflective surfaces of cars. Background regions of unbounded scenes encode inherent ambiguity of rendering, and specular reflections on cars violates the multi-view consistency. NeRF++ struggles in these scenes because of the enormous ambiguity. To deal with the challenges of rendering unbounded scenes with cars, we present a novel module to strengthen the capability of the basic model in this task. We propose to learn the positional bias between sampled points along a camera ray and target points along the incident light by multi-layer perceptrons to reconstitute the input points and view direction with regularization constraints for physical rendering. Considering the variety of materials and textures in unbounded scenes, we implicitly separate learned foreground colors into two components, diffuse and specular colors, to acquire smooth results. Our module improves basic models by 2.5% on average SSIM in our extensive experiments, produces more photo-realistic novel views of real-world unbounded scenes than other compared methods, and achieves the physical color editing of cars. [ABSTRACT FROM AUTHOR]

This paper presents a new direct adaptive control (DAC) technique using Caputo's definition of the fractional-order derivative. This is the first time a fractional-order adaptive law is introduced to work together with an integer-order stable manifold for approximating the uncertainty of a class of nonlinear systems. The DAC approach uses universal function approximators such as multi-layer perceptrons with one hidden layer or fuzzy systems to approximate the controller. This paper presents a new lemma, which elucidates and clarifies the link between the Caputo and the Riemann–Liouville definitions. The introduced lemma is useful in developing a Lyapunov candidate to prove the stability of using the proposed fractional-order adaptive law. This is further explained by a numerical example, which is provided to elucidate the practicality of using the fractional-order derivative for updating the approximator parameters. The main novelty of the results in this paper is a rigorous stability proof of the fractional DAC approach for a class of nonlinear systems that is subjected to unstructured uncertainty and deals with the adaptation mechanism using a traditional integer-order stable manifold. This makes the control scheme easier to implement in practice. The fractional-order adaptation law provides greater degrees of freedom and a potentially larger functional control structure than the conventional adaptive control. Finally, the paper demonstrates that traditional integer-order DAC is a special case of the more general fractional-order DAC scheme introduced here. [ABSTRACT FROM AUTHOR]

In comparison to convolutional neural networks (CNN), the newly created vision transformer (ViT) has demonstrated impressive outcomes in human pose estimation (HPE). However, (1) there is a quadratic rise in complexity with respect to image size, which causes the traditional ViT to be unsuitable for scaling, and (2) the attention process at the transformer encoder as well as decoder also adds substantial computational costs to the detector's overall processing time. Motivated by this, we propose a novel Going shallow and deeper with vIsion Transformers for human Pose estimation (GITPose) without CNN backbones for feature extraction. In particular, we introduce a hierarchical transformer in which we utilize multilayer perceptrons to encode the richest local feature tokens in the initial phases (i.e., shallow), whereas self-attention modules are employed to encode long-term relationships in the deeper layers (i.e., deeper), and a decoder for keypoint detection. In addition, we offer a learnable deformable token association module (DTA) to non-uniformly and dynamically combine informative keypoint tokens. Comprehensive evaluation and testing on the COCO and MPII benchmark datasets reveal that GITPose achieves a competitive average precision (AP) on pose estimation compared to its state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Accurate prediction of geological formation tops is a crucial task for optimizing hydrocarbon exploration and production activities. This research investigates and conducts a comprehensive comparative analysis of several advanced machine learning approaches tailored for the critical application of geological formation top prediction within the complex Norwegian Continental Shelf (NCS) region. The study evaluates and benchmarks the performance of four prominent machine learning models: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest ensemble method, and Multi-Layer Perceptron (MLP) neural network. To facilitate a rigorous assessment, the models are extensively evaluated across two distinct datasets - a dedicated test dataset and a blind dataset independent for validation. The evaluation criteria revolve around quantifying the models' predictive accuracy in successfully classifying multiple geological formation top types. Additionally, the study employs the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm as a baseline benchmarking technique to contextualize the relative performance of the machine learning models against a conventional clustering approach. Leveraging two model-agnostic feature importance analysis techniques - Permutation Feature Importance (PFI) and Shapley Additive exPlanations (SHAP), the investigation identifies and ranks the most influential input variables driving the predictive capabilities of the models. The comprehensive analysis unveils the MLP neural network model as the top-performing approach, achieving remarkable predictive accuracy with a perfect score of 0.99 on the blind validation dataset, surpassing the other machine learning techniques as well as the DBSCAN benchmark. However, the SVM model attains superior performance on the initial test dataset, with an accuracy of 0.99. Intriguingly, the PFI and SHAP analyses converge in consistently pinpointing depth (DEPT), revolution per minute (RPM), and Hook-load (HKLD) as the three most impactful parameters influencing model predictions across the different algorithms. These findings underscore the potential of sophisticated machine learning methodologies, particularly neural network-based models, to significantly enhance the accuracy of geological formation top prediction within the geologically complex NCS region. However, the study emphasizes the necessity for further extensive testing on larger datasets to validate the generalizability of the high performance observed. Overall, this research delivers an exhaustive comparative evaluation of state-of-the-art machine learning techniques, offering critical insights to guide the optimal selection, development, and real-world deployment of accurate and reliable predictive modeling strategies tailored for hydrocarbon exploration and reservoir characterization endeavors in the NCS. [ABSTRACT FROM AUTHOR]

Proactive analysis of patient pathways helps healthcare providers anticipate treatment-related risks, identify outcomes, and allocate resources. Machine learning (ML) can leverage a patient's complete health history to make informed decisions about future events. However, previous work has mostly relied on so-called black-box models, which are unintelligible to humans, making it difficult for clinicians to apply such models. Our work introduces PatWay-Net, an ML framework designed for interpretable predictions of admission to the intensive care unit (ICU) for patients with symptoms of sepsis. We propose a novel type of recurrent neural network and combine it with multi-layer perceptrons to process the patient pathways and produce predictive yet interpretable results. We demonstrate its utility through a comprehensive dashboard that visualizes patient health trajectories, predictive outcomes, and associated risks. Our evaluation includes both predictive performance – where PatWay-Net outperforms standard models such as decision trees, random forests, and gradient-boosted decision trees – and clinical utility, validated through structured interviews with clinicians. By providing improved predictive accuracy along with interpretable and actionable insights, PatWay-Net serves as a valuable tool for healthcare decision support in the critical case of patients with symptoms of sepsis. [ABSTRACT FROM AUTHOR]

Improving feature representations is a crucial task in object re-identification (Re-ID). Enhancement of discriminative features and suppression of irrelevant features have become standard approaches through the use of attentions. Long-range dependencies and more comprehensive image information can be considered by self-attentions. Additionally, the importance of pixels can be calibrated by multi-layer perceptrons. In this paper, Double-Layer Attention network for object re-identification (DLAReID) is proposed. Self-attentions and multi-layer perceptrons are combined in the convolutional neural network, which are the main structures in transformers. The limitations of capturing local dependencies in convolutional neural networks and the excessive parameters in transformers are addressed. 3×3 filters are replaced by the proposed double-layer self-attentions in the bottleneck blocks of the last layer of the network. The ability to capture global information is enhanced by the cooperation of the double-branch self-attention of the outer layer and the attention acting on the relative position encoding module of the inner layer. Good performance is achieved by the proposed method, as demonstrated by comprehensive experiments on multiple datasets. [ABSTRACT FROM AUTHOR]

Purpose: This study assesses the Multilayer Perceptron (MLP) neural network, complemented by other Machine Learning techniques (CART, PCA), in predicting the antimicrobial activity of 140 newly designed imidazolium chlorides against Klebsiella pneumoniae before synthesis. Emphasis is on leveraging molecular properties for predictive analysis. Methods: Classification and regression decision trees (CART) identified the top 200 predictive molecular descriptors. Principal Component Analysis (PCA) reduced these descriptors to 5 components, retaining 99.57% of raw data information. Antimicrobial activity, categorized as high or low, was based on experimentally proven minimal inhibitory concentration (MIC), with a cut-point at MIC = 0.856 mol/L. A 12-fold cross-validation trained the MLP (architecture 5-12-2 with 5 Principal Components). Results: The MLP exhibited commendable performance, achieving almost 90% correct classifications across learning, validation, and test sets, outperforming models without PCA dimension reduction. Key metrics, including accuracy (0.907), sensitivity (0.905), specificity (0.909), and precision (0.891), were notably high. These results highlight the MLP model's efficacy with PCA as a high-quality classifier for determining antimicrobial activity. Conclusions: The study concludes that the MLP neural network, along with CART and PCA, is a robust tool for predicting the antimicrobial activity class of imidazolium chlorides against Klebsiella pneumoniae. CART and PCA, used in this study, allowed input variable reduction without significant information loss. High classification accuracy and associated metrics affirm the method's potential utility in pre-synthesis assessments, offering valuable insights for antimicrobial compound design. [ABSTRACT FROM AUTHOR]

This paper presents novel way to achieve fast and accurate Artificial Neural Network (ANN) modeling of Radio Frequency (RF) front end Low Noise Amplifier (LNA). Multilayer perceptron neural network is implemented to model high gain, ultra-low noise pseudomorhic High Electron Mobility Transistor (pHEMT) based LNA. Datasets are developed for input and output that is obtained from the Electro-Magnetic simulator. Different neural networks such as PatternNet, FitNet, and CascadeForwardNet with different algorithms are trained, tested, and compared to learn the behavior of the amplifier, and the best model is analyzed. Neural networks are modeled for LNA S-parameters and NF. It is observed that greater than 99% accuracy is achieved for PatternNet Bayesian regularization algorithm with less number of hidden layers and also it is found that the simulation results are almost similar to the developed ANN. [ABSTRACT FROM AUTHOR]

Precipitation is a fundamental factor affecting many fields, including freshwater reservation, flood warning and prevention, agriculture, and hydropower planning. Observation-based precipitation data usually come from two primary sources, namely gauges and satellite images. The former provides high reliability but sparse coverage, while the latter offers fine-grained data but is still inaccurate. There have been several efforts to estimate precipitation, including mathematical and probabilistic models and machine learning-based techniques. All existing solutions consider the target problem as a satellite data calibration with gauge data as complementary information. However, this approach fails to provide accurate predictive results due to the sparsity of gauges and significant errors in satellite data. This paper presents a novel precipitation estimating method that highlights the importance of gauge data, the most trustworthy data source. To be more precise, we formulate the precipitation estimation issue as a spatial prediction task with the goal of predicting rainfall data for non-monitoring locations using gauge data at the monitored sites. To this end, we propose a data-driven approach that exploits the encoder–decoder architecture, graph neural network, and the multimodal data fusion strategy. Specifically, we design an encoder that leverages a graph neural network for capturing the spatial relationship among the gauges. Meanwhile, the decoder exploits the convolutional networks to learn the temporal correlation within the historical data. Finally, we integrate satellite images and meteorological information using a multimodal data fusion based on a multilayer perceptron to enhance prediction accuracy. The experimental results show that our proposed model increases the estimation accuracy from 24.3 to 65.2% compared to the state-of-the-art. [ABSTRACT FROM AUTHOR]

Awareness, short-term memory, and long-term memory are interrelated cognitive abilities that influence orthographic acquisition under Individual Differences. Connectionists ignore the role of biological grammar in language acquisition and consider external inputs or interventions as factors that shape abstract grammar through network mapping architecture. As new intervention methods, the use of computers in connectionist modeling of language acquisition has increasingly been developed through Artificial Neural Networks. New optimized and accurate computational models have not been applied to investigate foreign language orthography acquisition. The purpose of the present study is to use Multi-Layer Perceptron Neural Network-Gray Wolf Optimizer computational model to simulate and evaluate individualized cognitive abilities including orthographic awareness, orthographic short-term memory-related, and orthographic long-term memory-related abilities of Iranian English as Foreign Language learners in orthography acquisition. Eighteen Iranian learners were randomly selected and non-randomly assigned into three low-labeled, mean-labeled, and high-labeled groups for the experiment. In a mixed-methods design, three Multi-Layer Perceptron Neural Network-Gray Wolf Optimizers were established to solve the mapping problem i.e., English plural noun variation orthography. ANOVA was used for analyzing Mean Squared Errors and content analysis was used for interpreting interviews and observations. Results indicated that not only the orthographic awareness and short-term memory-related abilities are more important in predicting acquisition competence than the orthographic long-term memory-related ability but also they give compensatory support to long-term memory-related ability. The proposed valid and reliable model can be used to generate hypotheses related to Individual Differences. Findings are discussed and relevant implications are drawn for teachers and practitioners. [ABSTRACT FROM AUTHOR]

Post-training rounding, also known as quantization, of estimated parameters stands as a widely adopted technique for mitigating energy consumption and latency in machine learning models. This theoretical endeavor delves into the examination of the impact of rounding estimated parameters in key regression methods within the realms of statistics and machine learning. The proposed approach allows for the perturbation of parameters through an additive error with values within a specified interval. This method is elucidated through its application to linear regression and is subsequently extended to encompass radial basis function networks, multilayer perceptrons, regularization networks, and logistic regression, maintaining a consistent approach throughout. [ABSTRACT FROM AUTHOR]

In order to improve the control performance and reliability of the pulse-width modulation (PWM) rectifiers in electric vehicle (EV) charging systems, the evaluation of DC bus capacitor health status is critical. In order to accurately monitor the health status of DC bus capacitors, a data-driven model fusion method is developed. In the method, multi-layer perceptron, random forest, and XGBoost are adopted as the base learners that produce separate row predictions. The second-level learner, support vector machine (SVM) accepts the outputs of the previous learners and integrates them into the final health status prediction. Meanwhile, the feature vector is constructed by only collecting the grid voltage, the grid current and the AC component of DC bus voltage. With the feature vector as input, the proposed method is able to accurately predict the health status of DC bus capacitor. Finally, we built a three-phase PWM rectifier as an experimental platform for validation. The experimental results verify that the proposed method fully utilizes the advantages of data-driven and model fusion, and achieves a high accuracy in the health state evaluation of DC bus capacitor. [ABSTRACT FROM AUTHOR]

The challenge of maintaining consistent quality in injection molding is critical, yet conducting a comprehensive inspection is both costly and time consuming. Leveraging artificial intelligence, this study proposed using machine learning—specifically multilayer perception (MLP) models—to predict the quality of injection-molded parts. The accuracy of this approach largely relies on hyperparameter tuning, a process that can be cumbersome and suboptimal if performed through trial and error. The Taguchi method has the advantages of robustness, efficiency, and simplicity, and is a widely used robust optimization tool. However, this method assumes a linear relationship between factors, which limits the processing of complex systems where interactions between factors are nonlinear. Furthermore, the Taguchi method is sensitive to initial assumptions about factors and their levels, and the results may not reflect the true behavior of the system. To address this, a two-stage design-of-experiments method was devised that systematically identifies the optimal hyperparameter settings, including the maximum number of epochs, learning rate, momentum, activation function, minimum batch size, and numbers of hidden layers and nodes. The method is executed in two stages: (1) an L12 (21 × 35) orthogonal array is used to identify the primary factors affecting model accuracy and (2) an L8 (23) full-factorial experiment is conducted discover the combinations that yield the highest performance. Two experimental case studies, integrated circuit (IC) tray width prediction and optical component weight prediction, were used to validate the proposed method. The results revealed that the best hyperparameter settings resulted in validation and test accuracy of 96.83% and 95.30%, respectively, for IC tray width prediction. The average root-mean-square errors are 0.019 and 0.022 in model validation and test, respectively, for optical component weight prediction, with short computational time. The proposed method demonstrates how the systematic optimization of hyperparameters for MLP model can enhance the efficiency and stability of model training and can be used to advance quality control in the field of injection molding. [ABSTRACT FROM AUTHOR]

When solar irradiation is uniform along with the array, the P–V curve represents a unique maximum power point (MPP). If the cells undergo shade conditions in the presence of bypass diodes, the solar array's power is decreased, and the P–V curve of the array represents multiple local MPPs (LMPP) and a global MPP (GMPP). LMPPs might mislead the maximum power point tracking (MPPT) algorithms because their characteristics are identical to the MPP. Various studies have been conducted on partial shading conditions. This study uses parallel distributed maximum power point tracking (DMPPT) due to the advantages of this structure. A high-gain converter is presented to resolve the high conversion gain required by the DC/DC converter in this structure. This study also presents MLP and RBF networks for MPP tracking and compares their efficiencies under the same irradiation and partial shading conditions. Since determining optimal weight coefficients in MLP neural networks and variances, means, and weights in RBF networks play an essential role in their performance, this study uses four optimization algorithms of particle swarm optimization (PSO), gray wolf optimization (GWO), grasshopper optimization algorithm (GOA), and Harris Hawks optimization (HHO). Finally, an adaptive fuzzy-PID controller controls the three-phase grid-connected inverter. A comparison of the results shows that the efficiency of MLP and RBFII is almost the same, about 98–99%. Moreover, the accuracy of MLP networks is higher than RBF, and RBF networks' only advantage is shorter training time. In addition, RBF networks require much more activation functions for proper performance. The simulation outcomes confirm the superior efficacy of the HHO algorithm in training neural networks when compared to alternative algorithms. [ABSTRACT FROM AUTHOR]

Pre-training graph neural networks (GNNs) have been proposed to promote graph-related downstream tasks, such as link prediction and node classification. Most existing works employ contrastive learning to explore graph characteristics by enforcing positive sample pairs to be close and negative sample pairs to be distant after performing data augmentation on the input graph. However, these methods apply random operations on input data in data augmentation and sample pairs construction, which leads to neglecting central nodes and the neighbor relationship between nodes. To address the corresponding problem, we propose a novel framework for pre-training GNNs, named Neighborhood-Enhanced Contrast for Pre-Training Graph Neural Networks (NECPT). Specifically, we propose data augmentation strategy based on node centrality to preserve central nodes and corresponding edges, which is used to generate two semantically similar views from input graph. Notably, our NECPT constructs sample pairs by integrating the potential node neighbors in graph structure and semantic space to explore general graph regularities. After generating node representations with GNN encoders and multilayer perceptrons, contrastive sample pairs are selected from different node neighbors, which combines diverse neighborhood relations into contrastive learning. Finally, node representations obtained from the model are used to predict the attributes of nodes and edges, which extracts deep semantic connections between attribute and structure information. Extensive experiments on benchmark datasets in biology and chemistry demonstrate the effectiveness of our proposed approach. [ABSTRACT FROM AUTHOR]

In this study, we compare the classification accuracy achievable with linear support vector machine (L-SVM), K-nearest neighbor (KNN), and multilayer perceptron (MLP) methods for a multi-class EEG signal. This can be done in three phases. In phase one, band-pass filtering is applied to raw electroencephalogram (EEG) signals to decompose into five different frequency subbands. In phase two, we extract 10 important features from each subband. In phase three, these extracted feature sets are used as input to L-SVM, KNN, and MLP classifiers which categorize the sample data into three classes namely yoga, meditation, and combined yoga–meditation. Various performance measures for each classifier are evaluated and then compared to know which classifier is effective in the classification of the EEG data into yoga, meditation, and combined yoga–meditation groups. Performance measures such as confusion matrix, accuracy, sensitivity, specificity, precision, and F1 score are used to validate the performance of classifiers. Kruskal–Wallis test has been conducted to compare the classification performance of the linear SVM, KNN, and MLP classifier models. By comparing the classification accuracy between the three classifiers, L-SVM achieved the highest accuracy of 91.67%. [ABSTRACT FROM AUTHOR]

Chemical vapor deposition was used to produce multi-walled carbon nanotubes (MWCNTs), which were modified by Fe–Ni/AC catalysts to enhance CO2 adsorption. In this study, a new realm of possibilities and potential advancements in CO2 capture technology is unveiled through the unique combination of cutting-edge modeling techniques and utilization of the recently synthesized Fe–Ni/AC catalyst adsorbent. SEM, BET, and FTIR were used to analyze their structure and morphology. The surface area of MWCNT was found to be 240 m2/g, but after modification, it was reduced to 11 m2/g. The modified MWCNT showed increased adsorption capacity with higher pressure and lower temperature, due to the introduction of new adsorption sites and favorable interactions at lower temperatures. At 25 °C and 10 bar, it reached a maximum adsorption capacity of 424.08 mg/g. The optimal values of the pressure, time, and temperature parameters were achieved at 7 bar, 2646 S and 313 K. The Freundlich and Hill models had the highest correlation with the experimental data. The Second-Order and Fractional Order kinetic models fit the adsorption results well. The adsorption process was found to be exothermic and spontaneous. The modified MWCNT has the potential for efficient gas adsorption in fields like gas storage or separation. The regenerated M-MWCNT adsorbent demonstrated the ability to be reused multiple times for the CO2 adsorption process, as evidenced by the study. In this study, a feed-forward MLP artificial neural network model was created using a back-propagation training approach to predict CO2 adsorption. The most suitable and efficient MLP network structure, selected for optimization, consisted of two hidden layers with 25 and 10 neurons, respectively. This network was trained using the Levenberg–Marquardt backpropagation algorithm. An MLP artificial neural network model was created, with a minimum MSE performance of 0.0004247 and an R2 value of 0.99904, indicating its accuracy. The experiment also utilized the blank spreadsheet design within the framework of response surface methodology to predict CO2 adsorption. The proximity between the Predicted R2 value of 0.8899 and the Adjusted R2 value of 0.9016, with a difference of less than 0.2, indicates a high level of similarity. This suggests that the model is exceptionally reliable in its ability to predict future observations, highlighting its robustness. [ABSTRACT FROM AUTHOR]

As an essential hydrological variable, the river water level is vital in research fields such as agriculture. This study proposes a new multi-step river water level prediction method based on spatiotemporal correlations ST-ESMC-AE-XGBoost. First, the temporal correlation prediction models are established by conducting Pearson correlation analysis for different hydrological monitoring points: T-Models; secondly, ChebNet, a graph neural networks (GNN) based on the spatial topology of each hydrological monitoring point, is established from the perspective of spatial correlation; finally, AE-XGBoost is used to reconstruct spatiotemporal features and perform integrated prediction. This study uses the research object of the nine hydrological monitoring stations in the Yangtze River Basin. The experiments show that: (1) classic methods such as extreme learning machines (ELM) still have stable application value in hydrological prediction; (2) T-Models can further improve the performance of single models such as support vector machine (SVM), ELM, and multi-layer perceptrons (MLP) in multi-step water level prediction; (3) using the ChebNet can learn beneficial information from the geospatial structure of hydrological monitoring points, further improve the prediction effect of the spatiotemporal association integration model; (4) the multi-step water level prediction method ST-ESMC-AE-XGBoost proposed in this study has the highest prediction accuracy and the best generalization performance among all compared models. [ABSTRACT FROM AUTHOR]

We present SinGRAV, an attempt to learn a generative radiance volume from multi-view observations of a single natural scene, in stark contrast to existing category-level 3D generative models that learn from images of many object-centric scenes. Inspired by SinGAN, we also learn the internal distribution of the input scene, which necessitates our key designs w.r.t. the scene representation and network architecture. Unlike popular multi-layer perceptrons (MLP)-based architectures, we particularly employ convolutional generators and discriminators, which inherently possess spatial locality bias, to operate over voxelized volumes for learning the internal distribution over a plethora of overlapping regions. On the other hand, localizing the adversarial generators and discriminators over confined areas with limited receptive fields easily leads to highly implausible geometric structures in the spatial. Our remedy is to use spatial inductive bias and joint discrimination on geometric clues in the form of 2D depth maps. This strategy is effective in improving spatial arrangement while incurring negligible additional computational cost. Experimental results demonstrate the ability of SinGRAV in generating plausible and diverse variations from a single scene, the merits of SinGRAV over state-of-the-art generative neural scene models, and the versatility of SinGRAV by its use in a variety of applications. Code and data will be released to facilitate further research. [ABSTRACT FROM AUTHOR]

Purpose: Damage detection of civil infrastructures is vital in structural health monitoring; however, traditional methods are time-consuming and sometimes require skilled experts. With the development of machine learning and deep learning techniques, different artificial neural networks are being used to detect structural damages. However, individual networks can have errors in detecting damages and estimating the accuracy of predictions. Thus, ensemble learning techniques play an essential role in providing more reliable results by combining results obtained by individual models. In this study, a novel deep ensemble learning approach is proposed based on the stacked generalization method for detecting damages in the case of structural health monitoring. A distinguishing feature of the proposed approach is the utilization of a multi-headed deep artificial neural network architecture for classification problems, with a neural network as a meta-learner. Method: In the first step, the CNN, CAE, and LSTM models are trained separately to use the latter as input data for the meta-learner. Then, a deep multi-headed MLP neural network is designed as a meta-learner to combine predictions using the stacking ensemble method to enhance the result accuracy. Results: The proposed method is verified by three structures prepared for structural health monitoring validation. The results represent the promising scope of stacking ensemble learning applications for structural health monitoring and damage detection. Conclusion: The findings indicate that employing ensemble learning is more precise than a neural network when it comes to identifying structural damages. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has affected the health, economy, and all aspects of human lives around the world. Accurate prediction of the daily new cases of COVID-19 is critical for precise programming and taking needed measures to prevent the outbreak of it. Hence, in the present study, a new hybrid intelligent model is developed by hybridizing the artificial neural network (ANN) with the honey badger algorithm (HBA) to accurately predict the daily new cases of COVID-19 in Brazil, India, Russia, and the US. The performance of the hybrid model, namely HBA-ANN, was compared with the stand-alone ANN and gene expression programming (GEP) model using statistical criteria, such as correlation coefficient (R), root mean square error, scatter index (SI), and Nash–Sutcliffe efficiency (NSE), and graphical criteria, such as Taylor diagram, scatter plot, and box plot. According to the Taylor diagram, in each data series, the predicted values with the HBA-ANN model have the lowest distance from the observation points, indicating the high accuracy of the HBA-ANN model compared to the ANN and GEP models. Furthermore, the HBA-ANN model has high values for R (0.999, 0.899, 0.853, and 0.993 for Brazil, Russia, India, and the US, respectively) and could reduce the errors of the ANN model by 97.22, 43.68, 85.41, and 93.83% for Brazil, Russia, India, and the US, respectively. For each data series, the SI for the HBA-ANN model is less than 0.1, and the NSE is between 0 and 1, indicating a very good performance in predicting the COVID-19 cases with the developed model. Based on the superior performance of the HBA-ANN model, it is recommended to implement the HBA algorithm to increase the prediction accuracy of the models in medicine and other scientific fields. Moreover, it is suggested to investigate the accuracy of the hybridized HBA algorithm with other artificial intelligence (AI) models. [ABSTRACT FROM AUTHOR]

With the opening of the Stock Connect programs, the mainland China and Hong Kong stock markets are becoming more closely linked. In this paper, we develop a China's stock market risk early warning system. The proposed early warning system consists of three components. First, we use value at risk (VaR) to identify the stock market risk in which stock market risk is divided into multiple categories instead of two categories. Second, we construct a comprehensive indicator system in which basic indicators, technical indicators, overseas return rate indicators, and macroeconomic indicators are considered simultaneously. Third, we use four machine learning models, namely long short-term memory (LSTM), gate recurrent unit (GRU), multilayer perceptron (MLP), and EXtreme Gradient Boosting algorithm (XGBoost), to predict China's stock market risk. Experimental results show that: (1) Considering the macroeconomic indicators and basic indicators of Shanghai Composite Index (SSEC), ShenZhen Component Index (SZCZ) and Hang Seng Index (HSI) can significantly improve the performance of predicting China's stock market risk. (2) The opening of SH-HK Stock Connect program improves the predictive performance, but the opening of SZ-HK Stock Connect program decreases the predictive performance. (3) The indicators related to Hong Kong become more important after the SZ-HK Stock Connect program. [ABSTRACT FROM AUTHOR]

In this paper, a multilayer perceptron (MLP)-type artificial neural network model with a back-propagation training algorithm is utilized to model the bubble growth and bubble dynamics parameters in nucleate boiling with a non-uniform electric field. The influences of the electric field on different parameters that describe bubble's behaviors including bubble waiting time, bubble departure frequency, bubble growth time, and bubble departure diameter are considered. This study models single bubble dynamic behaviors of R113 created on a heater in an inconsistent electric field by utilizing a MLP neural network optimized by four different swarm-based optimization algorithms, namely: Salp Swarm Algorithm (SSA), Grey Wolf Optimizer (GWO), Artificial Bee Colony (ABC) algorithm, and Particle Swarm Optimization (PSO). For evaluating the model effectiveness, the MSE value (Mean-Square Error) of the artificial neural network model with various optimization algorithms is measured and compared. The results suggest that the optimal networks in the two-hidden layer and three-hidden layer models for the bubble departure diameter improve MSE by 33.85% and 35.27%, respectively, when compared with the best response in the one-hidden layer model. Additionally, for bubble growth time, the networks with two hidden layers and three hidden layers have the 44.51% and 45.85% reduction in error, when compared with the network with one hidden layer, respectively. For the departure frequency, the error reduction in the two-layer and three-layer networks is 46.85% and 62.32%, respectively. For bubble waiting time, the best networks in the two hidden-layer and three hidden-layer models improve MSE by 52.44% and 62.27% compared with the best 1HL model response, respectively. Also, the two algorithms of SSA and GWO are able to compete well (comparable MSE) with the PSO and ABC algorithms. [ABSTRACT FROM AUTHOR]

Clarification question generation is used mainly to solve the potential missing product information problem in e-commerce. However, the problem of not being able to better understand contextual semantic information has arisen when studying the diversity and specificity of clarification questions. This further leads to imprecise prediction of question keywords. With the aim of solving the current constraints, we propose the Transformer Encoder of Keyword Predictor Keyword-Conditioned Network (TEKPCNet). First, the transformer encoder has been proven to perform well in predicting external question keywords by encoding context. Then, the Text Convolutional Neural Network (TextCNN) and multilayer perceptron are combined to enhance the input of the encoder-decoder model. The diversity and specificity of the clarification question generation process should be further enhanced. Finally, the decoding process is completed by combining the attention mechanism. The generated clarification questions were evaluated in terms of two datasets using both automatic assessment and human evaluation. The experimental results show that the proposed model outperforms the other models. [ABSTRACT FROM AUTHOR]

This study investigates importance sampling under the scheme of minibatch stochastic gradient descent, under which the contributions are twofold. First, theoretically, we develop a neat tilting formula, which can be regarded as a general device for asymptotically optimal importance sampling. Second, practically, guided by the formula, we present an effective algorithm for importance sampling which accounts for the effects of minibatches and leverages the Markovian property of the gradients between iterations. Experiments conducted on artificial data confirm that our algorithm consistently delivers superior performance in terms of variance reduction. Furthermore, experiments carried out on real-world data demonstrate that our method, when paired with relatively straightforward models like multilayer perceptron and convolutional neural networks, outperforms in terms of training loss and testing error. [ABSTRACT FROM AUTHOR]

Interpretability fosters trust when humans and artificial intelligence (AI) systems interact, and its value for neural networks in particular cannot be overstated. In this paper, we analyse the emergence and value of neural modularity as it relates to interpretability. We compare the modularity evolved through connectivity constraints in terms of network Q-scores and examine the interpretable qualities of the resultant networks with functional subset regression. The connectivity constraints compared here include those proposed by previous research as well as several novel variations formulated to express neuron input competition and connections per neuron variance. Networks were evolved using HyperNEAT on a free-form substrate. The results indicate that the connection costs successfully promote the evolution of neural modularity across a variety of tasks and show that the novel connection cost variations are competitive with previously explored connection costs. The interpretability assessment shows that while the evolved networks' interpretable qualities are task dependent, two of the compared connection costs deliver statistically different functional module overlap distributions. However, recovered subnetwork module accuracies remain low, highlighting the key points for future research. [ABSTRACT FROM AUTHOR]

Evaluation of water quality is essential for protecting both the environment and human wellbeing. There is a paucity of research on using machine learning for classification of groundwater used for irrigation with fewer input parameters and still getting satisfactory results, despite earlier studies exploring its application in evaluating water quality. Studies are required to determine the feasibility of using machine learning to classify groundwater used for irrigation using minimal input parameters. In this study, we developed machine learning models to simulate the Irrigation Water Quality Index (IWQI) and an economic model that used an optimal number of inputs with the highest possible accuracy. We utilized eight classification algorithms, including the LightGBM classifier, CatBoost, Extra Trees, Random Forest, Gradient Boosting classifiers, Support Vector Machines, Multi-Layer Perceptrons, and K-Nearest Neighbors Algorithm. Two scenarios were considered, the first using six inputs, including conductivity, chloride ( Cl - ), bicarbonate ( HCO 3 - ), sodium ( Na + ), calcium ( Ca 2 + ), and magnesium ( Mg 2 + ), and the second using three parameters, including total hardness (TH), chloride ( Cl - ), and sulfate ( SO 4 2 - ) that were selected based on the Mutual Information (MI) result. The models achieved satisfactory performance, with the LightGBM classifier as the best model, yielding a 91.08% F1 score using six inputs, and the Extra Trees classifier as the best model, yielding an 86.30% F1 score using three parameters. Our findings provide a valuable contribution to the development of accurate and efficient machine learning models for water quality evaluation. [ABSTRACT FROM AUTHOR]

The purpose of this study was to compare different machine learning models for predicting daily air quality index (AQI) and evaluating air quality grade (AQG). The study used publicly available data from 2014 to 2019 for six pollutants (PM10, PM2.5, NO2, SO2, CO, O3). Four models (random forest (RF), gradient boosting (GB), Lasso Regression (LASSO), and the Stacked Regressor) were used for predicting AQI, while six models (K-Nearest Neighbors (KNN), support vector machines (SVM), decision tree (DT), multilayer perceptron (MLP), random forest (RF), and the Stacked Classifier) were used for forecasting AQG. The individual models were evaluated using different statistical measures, such as R-squared (R2), root mean square error (RMSE), mean absolute error (MAE), accuracy score (ACC), Matthew's Correlation Coefficient (MCC), and F1 score. The study found that the stack model performed consistently across all metric scores for AQI prediction. The stack model had an R2 score of 0.973, RMSE of 7.568, and MAE of 4.596, outperforming LASSO, GB, and RF. This indicates that the stack model was able to minimize the weaknesses of the individual models and provide a more accurate prediction. For AQG, the stack model also performed better across all metric scores, with an ACC of 0.970, MCC of 0.960, and F1 of 0.970, outperforming MLP, KNN, SVM, DT, and RF. The study concluded that stacked generalization machine learning models can be used for forecasting air quality index and grade with high efficiency and precision, mitigating the concerns of overfitting against individual models. [ABSTRACT FROM AUTHOR]

The occurrence and regularity of macrofungal fruitbody formation are influenced by meteorological conditions; however, there is a scarcity of data about the use of machine-learning techniques to estimate their occurrence based on meteorological indicators. Therefore, we employed an artificial neural network (ANN) to forecast fruitbody occurrence in mycorrhizal species of Russula and Amanita, utilizing meteorological factors and validating the accuracy of the forecast of fruitbody formation. Fungal data were collected from two locations in Western Hungary between 2015 and 2020. The ANN was the commonly used algorithm for classification problems: feed-forward multilayer perceptrons with a backpropagation algorithm to estimate the binary (Yes/No) classification of fruitbody appearance in natural and undisturbed forests. The verification indices resulted in two outcomes: however, development is most often studied by genus level, we established a more successful, new model per species. Furthermore, the algorithm is able to successfully estimate fruitbody formations with medium to high accuracy (60–80%). Therefore, this work was the first to reliably utilise the ANN approach of estimating fruitbody occurrence based on meteorological parameters of mycorrhizal specified with an extended vegetation period. These findings can assist in field mycological investigations that utilize sporocarp occurrences to ascertain species abundance. [ABSTRACT FROM AUTHOR]

In this study, surface roughness is analyzed within a supervised pattern recognition framework using machine learning methods to present a robust technique for the quantitative determination of surface roughness of rocks. To reach this goal, rock surfaces are classified based on statistical, fractal, geostatistical, directional, and spectral features obtained from the surface profiles as well as the results of direct shear tests. In this way, after the calculation of the features in more than 9000 profiles collected from 84 natural rock samples, a representative vector containing representative features of the surface profiles was introduced for each surface, and joint roughness coefficient (JRC) was back-calculated from 84 direct shear tests on the rock samples. Then, principal component analysis and linear discriminant analysis of the representative vectors were performed in order to prepare the inputs for the classification. Multilayer perceptron (MLP) and radial basis function (RBF) artificial neural networks and support vector machine (SVM) were trained for the classification of rock surfaces. On the other hand, convolutional neural network (CNN) in which features are automatically extracted from the image data of representative surface profiles was the other method for the classification of rock surfaces. The comparison of the results shows RBF is a robust and reliable classifier yielding the least classification error among the investigated methods. Furthermore, comparing the results of machine learning methods with those of traditional equations proves that the SVM, MLP, and RBF yield superior performance. [ABSTRACT FROM AUTHOR]

Evaluation of land use and land cover (LULC) change is among vital tools used for tracking environmental health and proper resource management. Remote sensing data was used to determine LULC change in Bahi (Manyoni) Catchment (BMC) in central Tanzania. Landsat satellite images from Landsat 5 TM and Landsat 8 OLI/TIRS were used, and support vector machine (SVM) algorithm was applied to classify the features of BMC. The obtained kappa values were 0.74, 0.83 and 0.84 for LULC maps of 1985, 2005 and 2021, respectively, which indicates the degree of accuracy from produced being substantial to almost perfect. Classified maps along with geospatial, socio-economic and climatic drivers with sufficient explanatory power were incorporated into MLP-NN to produce transition potential maps. Transition maps were subsequently used in cellular automata (CA)-Markov chain model to predict future LULC for BMC in immediate-future (2035), mid-future (2055) and far-future (2085). The findings indicate BMC is expected to experience significant expansion of agricultural lands and built land from 31.89 to 50.16% and 1.48 to 9.1% from 2021 to 2085 at the expense of open woodland, shrubland and savanna grassland. Low-yield crop production, water scarcity and population growth were major driving forces for rapid expansion of agricultural lands and overall LULC in BMC. The findings are essential for understanding the impact of LULC on hydrological processes and offer insights for the internal drainage basin (IDB) board to make necessary measures to lessen the expected dramatic changes in LULC in the future while sustaining harmonious balance with livelihood activities. [ABSTRACT FROM AUTHOR]

Forecasting short-term solar radiation is crucial for many solar energy applications. Additionally, solar energy has a lower environmental impact than conventional sources like fossil fuels and can be used for investment purposes through the construction of large solar farm facilities. To test, evaluate, and compare various solar radiation models, short-term observations of meteorological, astronomical, computational, and geographical data were collected at two distinct locations from 2012 to 2015. In this study, seven machine learning models were employed: multi-layer perceptron (MLP), feedforward backpropagation algorithm (FFBP), autoregressive integrated moving average (ARIMA), linear regression (LR), radial basis function neural network (RBFNN), random forest (RF), and Gaussian process regression (GPR) models. These models were used to forecast hourly global solar radiation (GSR) using the aforementioned data as model input. The performance of the selected models' forecast accuracy was thoroughly examined by assessing it for a typical day, for four seasons, and under three sky conditions. The RF model can forecast GSR with satisfactory accuracy, and MLP and GPR models provide better accuracy than LR, FFBP, RBF, and ARIMA models. For example, the R2 value range of RF are 0.9621 for Tetuan site and 0.9534 for Tangier site, respectively. Meanwhile, RF, MLP, and GPR models under-forecast few high radiation values on clear days, which may due to the differences in training and testing data ranges and distributions of the sky conditions. Finally, the obtained result of this study indicate that the proposed RF model is a reliable alternative for short-term global solar radiation forecasting due to its high forecast accuracy. [ABSTRACT FROM AUTHOR]