Ograniczanie wyników
Czasopisma help
Autorzy help
Lata help
Preferencje help
Widoczny [Schowaj] Abstrakt
Liczba wyników

Znaleziono wyników: 47

Liczba wyników na stronie
first rewind previous Strona / 3 next fast forward last
Wyniki wyszukiwania
Wyszukiwano:
w słowach kluczowych:  MLP
help Sortuj według:

help Ogranicz wyniki do:
first rewind previous Strona / 3 next fast forward last
EN
Based on data from the National Disaster Management Agency, South Sumatra is one of the provinces with a reasonably large drought-affected area, totalling 8,853,691.009 ha. Drought is a hydrometeorological disaster, characterised by anomalous rainfall below normal levels. Reduced rainfall can lead to decreased soil moisture, reduced river flows, and a general scarcity of water, which limits availability of water both on the surface and in the soil. To anticipate and mitigate the impacts of drought, an accurate forecasting system is essential for effective disaster management and mitigation. This research focuses on forecasting drought using the standardised precipitation index (SPI) based on Long Short-Term Memory (LSTM) and Multilayer Perceptron (MLP) algorithms. It compares LSTM and MLP algorithms by integrating rainfall data from the FY-4A satellite and observational rain gauges, which are processed to generate SPI values. These data are employed to train and test MLP and LSTM models in predicting future drought conditions. The results indicate that drought can be effectively predicted using both MLP and LSTM. However, the MLP outperforms the LSTM, as reflected by a higher Nash-Sutcliffe efficiency (NSE) value, a lower error rate, and a predicted date trend that more closely aligns with actual observations.
EN
The main objective of the planned effort is to provide analytical analyses of current intrusion detection systems grounded on ML algorithms. Furthermore, examined in this work are the useful data sets and several techniques already in use to develop an effective IDS using single, hybrid, and ensemble machine learning algorithms. The approaches in the literature have then been investi-gated under several criteria to provide a clear road and direction for the next projects that will be successful. Nowadays, companies of all kinds include an intrusion detection system (IDS), which inhibits cybercrime to protect the network, resources, and private data. Many strategies have been suggested and implemented up till now to prevent uncivil behaviour. Since machine learning (ML) approaches are successful, the proposed approach applied several ML models for the intrusion detection system. The CIC IoT 2023 Dataset is the one applied in this paper, and a two-step process for Intrusion detection was proposed. Tested with several techniques including random forest, XGBoost, logistic regression, MLP model, and RNN. Following fine-tuning, the federated learning model using neural networks had the best accuracy—99.84%.
EN
Accurate diagnosis of Parkinson′s disease, especially in its early stages, can be a challenging task. The application of machine learning (ML) techniques has helped improve the diagnostic accuracy of Parkinson′s disease (PD) detection but integration of diagnostic features in ML models for the prediction of disease progression has remained an unexplored research avenue. In this research work, Long Short Term Memory (LSTM) was trained using diagnostic features on Parkinson patients speech signals, to predict the disease progression while a Multilayer Perceptron (MLP) was trained on the same diagnostic features to detect PD. Diagnostic features were selected using two well known feature selection methods named Relief F and Sequential Forward Selection method. The integration of feature selection methods in LSTM model has resulted in PD progression forecast with an accuracy of 88.7%. Further more, with the application of input diagnostic features on MLP, PD stage was accurately detected with an accuracy of 98.63%, precision of 97.64% and recall of 98.8% showing model robustness and efficiency for its potential application in health care.
EN
Objective: The classification of high-frequency oscillations (HFOs) in epileptic brain signals remains a challenging task. As promising biomarkers, HFOs can assist in identifying epileptogenic zones in drug-resistant epilepsy, thereby supporting presurgical decision-making and seizure monitoring. This study aims to develop a robust framework for deep feature extraction and classification frameworks to improve automated HFO detection from intracranial EEG (iEEG) data. The framework was evaluated using iEEG recordings from 20 patients, with 28 intervals per patient, each segmented using a 300 ms time window, resulting in a total of 900 trials. Methods: A deep-feature extraction system based on the GoogLeNet architecture was implemented to obtain discriminative representations of iEEG segments. The extracted deep features were classified using four well-known machine-learning algorithms: support vector machine (SVM), multilayer perceptron (MLP), Gaussian naïve Bayes (GNB) and random forest (RF). A suite of comparative experiments was conducted to evaluate the performance of the classification system and its robustness. Results: The proposed deep-feature extraction model, combined with the MLP classifier, achieved the best performance, with an overall accuracy of 95.18%, outperforming the other classifiers tested. Results confirm that deep features, extracted via GoogLeNet, significantly improve the precision of HFO classification compared to conventional featurebased methods. Conclusions: The proposed hybrid approach offers a reliable and efficient method for classifying HFOs in iEEG data. Its excellent performance demonstrates the potential for integration into clinical workflows to assist neurologists in identifying epileptogenic regions and monitoring seizures. This study contributes to advancing automated epilepsy diagnostics through the use of deep-learning-based feature extraction.
EN
Forecasting relative humidity is a critical for addressing the challenges of climate change. It facilitates comprehension of climatic mechanisms and the anticipation of extreme weather events, while also contributing to strengthening societal resilience and protection. Indeed, elevated levels of humidity have been demonstrated to exacerbate heat waves, leading to a marked increase in both the perceived temperature and the associated health risks. Conversely, low humidity promotes conditions conducive to droughts and wildfires. Moreover, relative humidity plays a key role in the water cycle, influencing precipitation, evaporation, and cloud formation. Understanding these mechanisms is essential for anticipating floods, droughts, and water shortages. In this study, mathematical models were developed to predict relative humidity in the Fez, Morocco, using multilayer perceptron (MLP) neural networks, radial basis function (RBF) neural networks, and multiple linear regression (MLR). The dataset used in this study includes daily values of eight meteorological parameters, including temperature at 2m, shortwave Radiation, diffuse shortwave radiation, precipitation total, evapotranspiration, vapor pressure deficit and wind speed and relative humidity as the output. The data spans 38 years, from January 1985 to December 2022, and includes 13879 observation days.. To evaluate the predictive performance of these models, we analyzed their architectures, learning algorithms, correlation coefficients, and mean squared errors. The results indicate that the MLP model attains the highest predictive accuracy, with a correlation coefficient of 0.9809 and a mean squared error MSE of 0.0099, outperforming the RBF model (correlation of 0.9603) and the MLR model (correlation of 0.9023), the best performing model used a Tansig activation function in the hidden layer, a Purelin function in the output layer and the Levenberg-Marquardt learning algorithm with a MLP configuration [7-15-1]. The findings of this study offer a valuable contribution to the field of water resource management in the region. They demonstrate the efficacy of artificial neural network models in enhancing moisture forecasting, thereby providing a solid foundation for future research in climate modelling.
6
Content available Recognizing user emotion based on keystroke dynamics
EN
The paper presents a study concerning recognizing user emotion based on keystroke dynamics of the written text. At first, the analysis of the dataset used in the task is performed. Followed by the training and the effectiveness assessment of classical methods: Naive Bayes, K-Nearest Neighbours, Random Forest, and Multilayer Perceptron applied to the classification of provided samples to one of four emotions: anger, calm, happiness, sadness. The precision, recall, F1-score and time performance are evaluated. The Random Forest and MLP classifiers performed best, with an overall F1 measure of 84.83% and 80.47%, respectively. The scenario for extending the data set is proposed, along with the analysis of classification results of new data.
PL
Artykuł przedstawia badania dotyczące rozpoznawania emocji użytkownika na podstawie dynamiki naciśnięć klawiszy wpisywanego tekstu. W pracy przeprowadzono analizę wykorzystywanego zbioru danych, wytrenowano oraz dokonano oceny skuteczności klasycznych metod takich jak: naiwny klasyfikator Bayesa, metoda najbliższych sąsiadów, las losowy oraz perceptron wielowarstwowy, zastosowanych do przyporządkowania danych do jednej z czterech emocji: złości, spokoju, radości lub smutku. Uzyskane wyniki zostały ewaluowane z wykorzystaniem miar precyzji, czułości oraz F1, oceniono również wydajność czasową. Las losowy oraz perceptron wielowarstwowy osiągnęły najlepsze wyniki, z wynikiem F1 równym odpowiednio 84.83% i 80.47%. Zaprezentowano również scenariusz rozszerzenia zbioru danych, razem z analizą wyników klasyfikacji nowych danych.
EN
Nowadays, the world is struggling with the problems of an aging society. With the increasing share of older people in the population, degenerative joint diseases are a growing problem. The result of progressive degenerative changes in joints is primarily the deterioration of the patients' quality of life and their gradual exclusion from activity and social life. The ability to effectively, non-invasively and quickly detect cases of chondromalacia of the knee joints is a challenge for modern medicine. The possibility of early detection of progressive degenerative changes allows for the appropriate selection of treatment protocols and significantly increases the chances of inhibiting the development of degenerative diseases of the musculoskeletal system. The article presents a non-invasive method for detecting degenerative changes in the knee joints based on recurrence analysis and classification using neural networks. The result of the analyzes was a classification accuracy of 91.07% in the case of MLP neural networks and 80.36% for RBF networks.
EN
Widespread proliferation of interconnected healthcare equipment, accompanying software, operating systems, and networks in the Internet of Medical Things (IoMT) raises the risk of security compromise as the bulk of IoMT devices are not built to withstand internet attacks. In this work, we have developed a cyber-attack and anomaly detection model based on recursive feature elimination (RFE) and multilayer perceptron (MLP). The RFE approach selected optimal features using logistic regression (LR) and extreme gradient boosting regression (XGBRegressor) kernel functions. MLP parameters were adjusted by using a hyperparameter optimization and 10-fold cross-validation approach was performed for performance evaluations. The developed model was performed on various IoMT cybersecurity datasets, and attained the best accuracy rates of 99.99%, 99.94%, 98.12%, and 96.2%, using Edith Cowan University- Internet of Health Things (ECU-IoHT), Intensive Care Unit (ICU Dataset), Telemetry data, Operating systems’ data, and Network data from the testbed IoT/IIoT network (TON-IoT), and Washington University in St. Louis enhanced healthcare monitoring system (WUSTL-EHMS) datasets, respectively. The proposed method has the ability to counter cyber attacks in healthcare applications.
EN
Automatic segmentation of breast lesions from ultrasound images plays an important role in computer-aided breast cancer diagnosis. Many deep learning methods based on convolutional neural networks (CNNs) have been proposed for breast ultrasound image segmentation. However, breast ultrasound image segmentation is still challenging due to ambiguous lesion boundaries. We propose a novel dual-stage framework based on Transformer and Multi-layer perceptron (MLP) for the segmentation of breast lesions. We combine the Swin Transformer block with an efficient pyramid squeezed attention block in a parallel design and introduce bi-directional interactions across branches, which can efficiently extract multi-scale long-range dependencies to improve the segmentation performance and robustness of the model. Furthermore, we introduce tokenized MLP block in the MLP stage to extract global contextual information while retaining fine-grained information to segment more complex breast lesions. We have conducted extensive experiments with state-of-the-art methods on three breast ultrasound datasets, including BUSI, BUL, and MT_BUS datasets. The dice coefficient reached 0.8127 ± 0.2178, and the intersection over union reached 0.7269 ± 0.2370 on benign lesions when the Hausdorff distance was maintained at 3.75 ± 1.83. The dice coefficient of malignant lesions is improved by 3.09% for BUSI dataset. The segmentation results on the BUL and MT_BUS datasets also show that our proposed model achieves better segmentation results than other methods. Moreover, the external experiments indicate that the proposed model provides better generalization capability for breast lesion segmentation. The dual-stage scheme and the proposed Transformer module achieve the fine-grained local information and long-range dependencies to relieve the burden of radiologists.
EN
Osteoarthritis is one of the most common cause of disability among elderly. It can affect every joint in human body, however, it is most prevalent in hip, knee, and hand joints. Early diagnosis of cartilage lesions is essential for fast and accurate treatment, which can prolong joint function. Available diagnostic methods include conventional X-ray, ultrasound and magnetic resonance imaging. However, those diagnostic modalities are not suitable for screening purposes. Vibroarthrography is proposed in literature as a screening method for cartilage lesions. However, exact method of signal acquisition as well as classification method is still not well established in literature. In this study, 84 patients were assessed, of whom 40 were in the control group and 44 in the study group. Cartilage status in the study group was evaluated during surgical treatment. Multilayer perceptron - MLP, radial basis function - RBF, support vector method - SVM and naive classifier – NBC were introduced in this study as classification protocols. Highest accuracy (0.893) was found when MLP was introduced, also RBF classification showed high sensitivity (0.822) and specificity (0.821). On the other hand, NBC showed lowest diagnostic accuracy reaching 0.702. In conclusion vibroarthrography presents a promising diagnostic modality for cartilage evaluation in clinical setting with the use of MLP and RBF classification methods.
EN
Water resources, consisting of surface water and groundwater, are considered to be among the crucial natural resources in most arid and semiarid regions. Groundwater resources as the sustainable yields can be predicted, whereas this is one of the important stages in water resource management. To this end, several models such as mathematical, statistical, empirical, and conceptual can be employed. In this paper, machine learning and deep learning methods as conceptual ones are applied for the simulations. The selected models are support vector regression (SVR), adaptive neuro-fuzzy inference system (ANFIS), and multilayer perceptron (MLP). Next, these models are optimized with the adaptive moment estimation (ADAM) optimization algorithm which results in hybrid models. The hyper-parameters of the stated models are optimized with the ADAM method. The root mean squared error (RMSE), mean absolute error (MAE), mean squared error (MSE), and coefficient of determination (R2) are used to evaluate the accuracy of the simulated groundwater level. To this end, the aquifer hydrograph is used to compare the results with observations data. So, the RMSE and R2 show that the accuracy of the machine learning and deep learning models is better than the numerical model for the given data. Moreover, the MSE is approximately the same in all three cases (ranging from 0.7113 to 0.6504). Also, the total value of R2 and RMSE for the best hybrid model is 0.9617 and 0.7313, respectively, which are obtained from the model output. The results show that all three techniques are useful tools for modeling hydrological processes in agriculture and their computational capabilities and memory are similar.
EN
This article investigates the application of neural network models to create automated control systems for industrial processes. We reviewed and analysed works on dispatch control and evaluation of equipment operating modes and the use of artificial neural networks to solve problems of this type. It is shown that the main requirements for identification models are the accuracy of estimation and ease of algorithm implementation. It is shown that artificial neural networks meet the requirements for accuracy of classification problems, ease of execution and speed. We considered the structures of neural networks that can be used to recognise the modes of operation of technological equipment. Application of the model and structure of networks with radial basis functions and multilayer perceptrons for identifying the mode of operation of equipment under given conditions is substantiated. The input conditions for constructing neural network models of two types with a given three-layer structure are offered. The results of training neural models on the model of a multilayer perceptron and a network with radial basis functions are presented. The estimation and comparative analysis of models depending on model parameters are made. It is shown that networks with radial basis functions offer greater accuracy in solving identification problems. The structural scheme of the automated process control system with mode identification based on artificial neural networks is offered.
EN
This article accounts for the development of a powerful artificial neural network (ANN) model, designed for the prediction of relative humidity levels, using other meteorological parameters such as the maximum temperature, minimum temperature, precipitation, wind speed, and intensity of solar radiation in the Rabat-Kenitra region (a coastal area where relative humidity is a real concern). The model was applied to a database containing a daily history of five meteorological parameters collected by nine stations covering this region from 1979 to mid-2014. It has been demonstrated that the best performing three-layer (input, hidden, and output) ANN mathematical model for the prediction of relative humidity in this region is the multi-layer perceptron (MLP) model. This neural model using the Levenberg-Marquard algorithm, with an architecture of [5-11-1] and the transfer functions Tansig in the hidden layer and Purelin in the output layer, was able to estimate relative humidity values that were very close to those observed. This was affirmed by a low mean squared error (MSE) and a high correlation coefficient (R), compared to the statistical indicators relating to the other models developed as part of this study.
EN
The paper evaluated the possibility of using artificial neural network models for predicting the compressive strength (Fc) of concretes with the addition of recycled concrete aggregate (RCA). The artificial neural network (ANN) approaches were used for three variable processes modeling (cement content in the range of 250 to 400 kg/m3, percentage of recycled concrete aggregate from 25% to 100% and the ratios of water contents 0.45 to 0.6). The results indicate that the compressive strength of recycled concrete at 3, 7 and 28 days is strongly influenced by the cement content, %RCA and the ratios of water contents. It is found that the compressive strength at 3, 7 and 28 days decreases when increasing RCA from 25% to 100%. The obtained MLP and RBF networks are characterized by satisfactory capacity for prediction of the compressive strength of concretes with recycled concrete aggregate (RCA) addition. The results in statistical terms; correlation coefficient (R) reveals that the both ANN approaches are powerful tools for the prediction of the compressive strength.
EN
This research paper investigates the application of neural network models for forecasting in energy. The results of forecasting the weekly energy consumption of the enterprise according to the model of a multilayer perceptron at different values of neurons and training algorithms are given. The estimation and comparative analysis of models depending on model parameters is made.
16
Content available remote Neural network model for enterprise energy consumption forecasting
EN
This research paper investigates the application of neural network models for forecasting in energy. The results of forecasting the weekly energy consumption of the enterprise according to the model of a multilayer perceptron at different values of neurons and training algorithms are given. The estimation and comparative analysis of models depending on model parameters is made.
17
EN
The quantitative analyses of karst spring discharge typically rely on physical-based models, which are inherently uncertain. To improve the understanding of the mechanism of spring discharge fuctuation and the relationship between precipitation and spring discharge, three machine learning methods were developed to reduce the predictive errors of physical-based groundwater models, simulate the discharge of Longzici spring’s karst area, and predict changes in the spring on the basis of long time series precipitation monitoring and spring water fow data from 1987 to 2018. The three machine learning methods included two artifcial neural networks (ANNs), namely multilayer perceptron (MLP) and long short-term memory–recurrent neural network (LSTM–RNN), and support vector regression (SVR). A normalization method was introduced for data preprocessing to make the three methods robust and computationally efcient. To compare and evaluate the capability of the three machine learning methods, the mean squared error (MSE), mean absolute error (MAE), and root-mean-square error (RMSE) were selected as the performance metrics for these methods. Simulations showed that MLP reduced MSE, MAE, and RMSE to 0.0010, 0.0254, and 0.0318, respectively. Meanwhile, LSTM–RNN reduced MSE to 0.0010, MAE to 0.0272, and RMSE to 0.0329. Moreover, the decrease in MSE, MAE, and RMSE was 0.0397, 0.1694, and 0.1991, respectively, for SVR. Results indicated that MLP performed slightly better than LSTM–RNN, and MLP and LSTM–RNN performed considerably better than SVR. Furthermore, ANNs were demonstrated to be prior machine learning methods for simulating and predicting karst spring discharge.
EN
Electricity demand forecasting is a term used for prediction of users’ consumption on the grid ahead of actual demand. It is very important to all power stakeholders across levels. The power players employ electricity demand forecasting for sundry purposes. Moreover, the government’s policy on its market deregulation has greatly amplified its essence. Despite numerous studies on the subject using certain classical approaches, there exists an opportunity for exploration of more sophisticated methods such as the deep learning (DL) techniques. Successful researches about DL applications to computer vision, speech recognition, and acoustic computing problems are motivation. However, such researches are not sufficiently exploited for electricity demand forecasting using DL methods. In this paper, we considered specific DL techniques (LSTM, CNN, and MLP) to short-term load forecasting problems, using tropical institutional data obtained from a Transmission Company. We also test how accurate are predictions across the techniques. Our results relatively revealed models appropriateness for the problem.
EN
High order modulation (HOM) presents a key challenge in increasing spectrum efficiency in 4G and upcoming 5G communication systems. In this paper, two non-linear adaptive equalizer techniques based on multilayer perceptron (MLP) and radial basis function (RBF) are designed and applied on HOM to optimize its performance despite its high sensitivity to noise and channel distortions. The artificial neural network’s (ANN) adaptive equalizer architectures and learning methods are simplified to avoid more complexity and to ensure greater speed in symbol decision making. They will be compared with the following popular adaptive filters: least mean square (LMS) and recursive least squares (RLS), in terms of bit error rate (BER) and minimum square error (MSE) with 16, 64, 128, 256, 512 and 1024 quadrature amplitude modulation (QAM). By that, this work will show the advantage that the MLP equalizer has, in most cases, over RBF and traditional linear equalizers.
EN
The aim of the following work is to indicate factors which significantly affect the emergence of selected soybean varieties after application of natural herbal extracts based on - Levisticum officinale L., Ribes nigrum L., Matricaria chamomilla L., as wet seed treatments using two methods of treatment. The research material included seeds treated for 24 hours in macerats, decoctions and infusions made from the above herb species as well as untreated seeds, seeded together with preparations in point application. Untreated seeds were used as the control group. The experiment was being conducted for 16 days in a greenhouse facility belonging to the COBORU Experimental Station for Variety Testing in Karzniczka. The assessed parameter referred to the percentage of soybean seedlings emergence ability determined based on the number of emerged plants. Indication of the importance of factors in shaping soybean emergence and considering their rank was possible due to the sensitivity analysis of the generated neural network with the MLP architecture 4:4-13-5-1:1 with two hidden layers. All analyzed factors of the experiment significantly shaped the ability of soybean emergence, with the following order: cultivar, application method, herb species from which the extract was made, form of preparation.
PL
Celem pracy było wskazanie czynników istotnie wpływających na wschody wybranych odmian soi po zastosowaniu naturalnych ekstraktów wodnych na bazie ziół - Levisticum officinale L., Ribes nigrum L., Matricaria chamomilla L., jako zapraw nasiennych na mokro z wykorzystaniem dwóch sposobów zaprawiania. Materiał badawczy stanowiły nasiona zaprawiane przez dobę w maceratach, wywarach i naparach sporządzonych z powyższych gatunków ziół oraz nasiona niezaprawiane, wysiewane łącznie z aplikacją punktową preparatów. Za obiekt kontrolny przyjęto nasiona niezaprawiane. Eksperyment prowadzono przez 16 dni w obiekcie szklarniowych należącym do Stacji Doświadczalnej Oceny Odmian COBORU w Karzniczce. Parametrem poddanym ocenie była procentowa zdolność wschodów siewek soi określana na podstawie liczby wzeszłych roślin. Wskazanie istotności czynników w kształtowaniu zdolności wschodów soi oraz uwzględnienie ich rangi było możliwe dzięki analizie wrażliwości wytworzonej sieci neuronowej o architekturze MLP 4:4-13-5-1:1 z dwoma ukrytymi warstwami. Wszystkie analizowane czynniki doświadczenia znacząco kształtowały zdolność wschodów soi, a ich waga miała następującą kolejność: odmiana, sposób aplikacji preparatu, gatunek zioła, z którego sporządzono ekstrakt, forma preparatu.
first rewind previous Strona / 3 next fast forward last
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.