Accuracy Comparison Between Feedforward Neural Network, Support Vector Machine and Boosting Ensembles for Financial Risk Evaluation DOI Open Access
Dat Tran,

Allan W. Tham

Journal of risk and financial management, Journal Year: 2025, Volume and Issue: 18(4), P. 215 - 215

Published: April 15, 2025

Loan defaults have become an increasing concern for lending institutions, presenting significant challenges to profitability and operational stability. However, with the advent of advanced data processing capabilities, greater availability, development sophisticated machine learning techniques—particularly neural networks—new opportunities emerged classifying predicting loan beyond traditional manual methods. This, in turn, can reduce risk enhance overall financial performance. In recent years, institutions increasingly employed these techniques mitigate non-performing loans (NPLs) by improving approval efficiency. This study aims address a gap literature examining predictive performance different network architectures on datasets. Specifically, it compares effectiveness Feedforward Neural Networks (FNNs), Long Short-Term Memory (LSTM) networks, one-dimensional Convolutional (1D-CNNs) forecasting defaults. Despite growing body research this area, comparative studies focusing application various default prediction remain relatively scarce.

Language: Английский

Deep learning based long-term global solar irradiance and temperature forecasting using time series with multi-step multivariate output DOI
Narjes Azizi, Maryam Yaghoubirad, Meisam Farajollahi

et al.

Renewable Energy, Journal Year: 2023, Volume and Issue: 206, P. 135 - 147

Published: Feb. 7, 2023

Language: Английский

Citations

40
Hydrological drought forecasting and monitoring system development using artificial neural network (ANN) in Ethiopia DOI Creative Commons
Kassa Abera Tareke,

Admasu Gebeyehu Awoke

Heliyon, Journal Year: 2023, Volume and Issue: 9(2), P. e13287 - e13287

Published: Jan. 29, 2023

The objective of this study is to investigate and perform long-term forecasting both streamflow hydrological drought over Ethiopia. Observed precipitation data are collected from 17 stations 34 rainfall gauge forecast future 2026 2099. Streamflow performed using an artificial neural network (ANN) in conjunction with python software. 1973 2014 used train test the ANN model by 70 30% ratios, respectively. After training model, downscaled regional climate models (RCM) have been as input streamflow. Three RCM were downscale historical data. RACMO found a good downscaling for all selected stations. linear scaling bias correction technique results less than 2% error compared other alternative techniques. result indicates that tool areas having correlation between such Abbay, Awash, Baro, Omo Gibe, Tekeze river basins. But arid example Genale Dawa, Wabishebele, Rift Valley basins, not suitable because (precipitation) high variation output variable (streamflow). In areas, meteorological analysis better analysis. Finally, analyzed forecasted index (SDI). 2028, 2036, 2042, 2044, 2062, 2063 expected extreme years most basins Ethiopia future. This shows at least one each decade Therefore, extensive research needed develop effective early warning system, water resource management policy.

Language: Английский

Citations

28
A Novel Smoothing-Based Deep Learning Time-Series Approach for Daily Suspended Sediment Load Prediction DOI
Bibhuti Bhusan Sahoo, Sovan Sankalp, Özgür Kişi

et al.

Water Resources Management, Journal Year: 2023, Volume and Issue: 37(11), P. 4271 - 4292

Published: July 22, 2023

Language: Английский

Citations

25
Advancements in Natural Language Processing: Implications, Challenges, and Future Directions DOI Creative Commons
Supriyono Supriyono, Aji Prasetya Wibawa,

Suyono

et al.

Telematics and Informatics Reports, Journal Year: 2024, Volume and Issue: 16, P. 100173 - 100173

Published: Nov. 7, 2024

Language: Английский

Citations

12
Multimodal 1D CNN for delamination prediction in CFRP drilling process with industrial robots DOI
Jae Gyeong Choi, Dong Chan Kim, M. Chung

et al.

Computers & Industrial Engineering, Journal Year: 2024, Volume and Issue: 190, P. 110074 - 110074

Published: March 13, 2024

Language: Английский

Citations

11
Comparative Analysis of Convolutional Neural Network-Long Short-Term Memory, Sparrow Search Algorithm-Backpropagation Neural Network, and Particle Swarm Optimization-Extreme Learning Machine Models for the Water Discharge of the Buzău River, Romania DOI Open Access
Zhen Liu, Alina Bărbulescu

Water, Journal Year: 2024, Volume and Issue: 16(2), P. 289 - 289

Published: Jan. 15, 2024

Modeling and forecasting the river flow is essential for management of water resources. In this study, we conduct a comprehensive comparative analysis different models built monthly discharge Buzău River (Romania), measured in upper part river’s basin from January 1955 to December 2010. They employ convolutional neural networks (CNNs) coupled with long short-term memory (LSTM) networks, named CNN-LSTM, sparrow search algorithm backpropagation (SSA-BP), particle swarm optimization extreme learning machines (PSO-ELM). These are evaluated based on various criteria, including computational efficiency, predictive accuracy, adaptability training sets. The obtained applying CNN-LSTM stand out as top performers, demonstrating superior efficiency high especially when set containing data series 1984 (putting Siriu Dam operation) September 2006 (Model type S2). This research provides valuable guidance selecting assessing prediction models, offering practical insights scientific community real-world applications. findings suggest that Model S2 preferred choice forecast predictions due its speed accuracy. S (considering recorded 2006) recommended secondary option. S1 (with period 1955–December 1983) suitable other unavailable. study advances field by presenting precise these their respective strengths

Language: Английский

Citations

9
A comprehensive benchmark of machine learning-based algorithms for medium-term electric vehicle charging demand prediction DOI Creative Commons
Ömer Can Tolun, Kasım Zor, Önder Tutsoy

et al.

The Journal of Supercomputing, Journal Year: 2025, Volume and Issue: 81(3)

Published: Feb. 10, 2025

Language: Английский

Citations

1
GBT: Two-stage transformer framework for non-stationary time series forecasting DOI Creative Commons
Li Shen, Yuning Wei,

Yangzhu Wang

et al.

Neural Networks, Journal Year: 2023, Volume and Issue: 165, P. 953 - 970

Published: July 5, 2023

This paper shows that time series forecasting Transformer (TSFT) suffers from severe over-fitting problem caused by improper initialization method of unknown decoder inputs, especially when handling non-stationary series. Based on this observation, we propose GBT, a novel two-stage framework with Good Beginning. It decouples the prediction process TSFT into two stages, including Auto-Regression stage and Self-Regression to tackle different statistical properties between input sequences. Prediction results serve as 'Good Beginning', i.e., better for inputs stage. We also Error Score Modification module further enhance capability in GBT. Extensive experiments seven benchmark datasets demonstrate GBT outperforms SOTA TSFTs (FEDformer, Pyraformer, ETSformer, etc.) many other models (SCINet, N-HiTS, only canonical attention convolution while owning less space complexity. is general enough couple these strengthen their capability. The source code available at: https://github.com/OrigamiSL/GBT.

Language: Английский

Citations

20
A shale gas production prediction model based on masked convolutional neural network DOI
Wei Zhou,

Xiangchengzhen Li,

ZhongLi Qi

et al.

Applied Energy, Journal Year: 2023, Volume and Issue: 353, P. 122092 - 122092

Published: Oct. 17, 2023

Language: Английский

Citations

20
EMDFormer model for time series forecasting DOI Creative Commons
Ana Lazcano, Miguel A. Jaramillo-Morán, Julio E. Sandubete

et al.

AIMS Mathematics, Journal Year: 2024, Volume and Issue: 9(4), P. 9419 - 9434

Published: Jan. 1, 2024

<abstract> <p>The adjusted precision of economic values is essential in the global economy. In recent years, researchers have increased their interest making accurate predictions this type time series; one reasons that characteristics series makes predicting a complicated task due to its non-linear nature. The evolution artificial neural network models enables us research suitability generated for other purposes, applying potential prediction with promising results. Specifically, field, application transformer assuming an innovative approach great To improve performance networks, work, empirical model decomposition (EMD) methodology was used as data preprocessing network. results confirmed better compared networks widely bidirectional long short term memory (BiLSTM), and (LSTM) using without EMD preprocessing, well comparison Transformer data, lower error all metrics used: root mean square (RMSE), (MSE), absolute percentage (MAPE), R-square (R<sup>2</sup>). Finding provides literature allows greater adjustment minimal preprocessing.</p> </abstract>

Language: Английский

Citations

6