GIS-based seismic vulnerability mapping: a comparison of artificial neural networks hybrid models

Yariyan, Peyman; Abbaspour, Rahim Ali; Chehreghan, Alireza; Karami, Mohammadreza; Cerdà, Artemi

doi:10.1080/10106049.2021.1892208

Cited by 11 publications

(9 citation statements)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the purpose of predicting and preventing earthquake damage, Earthquake Risk Mapping (ERM) is crucial knowledge. In order to construct hybrid models for ERM, Yariyan et al [36] study employed the Classification Tree Analysis (CTA) learner model with three Gini, Entropy, Ratio split, and Fuzzy ART MAP (FAM) models. The Earthquake Risk Conditioning Factors, which encompass environmental, physical, and social elements, were chosen based on the opinions and experiences of experts.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Prediction of earthquake magnitude and seismic vulnerability mapping using artificial intelligence techniques: a case study of Turkey

Biswas

Kumar

Bera

2023

Preprint

View full text Add to dashboard Cite

Earthquake threats can result in fatalities, property destruction, and other cascading effects. Since it is nearly impossible to prevent earthquakes, anticipating the location of future earthquakes and figuring out their likelihood could be very helpful in reducing the seismic threat. In this work, seismic hazard prediction is executed to forecast adverse results using a range of potential artificial intelligence (AI) techniques, including ML and ANN. In the case study, we have looked at Turkey, which was recently and badly damaged by two earthquakes in February 2023. To predict earthquake magnitude, this study used a variety of regression algorithms, including Decision Tree Regressor, Extra-Trees Regressor, Random Forest Regressor, Bayesian Ridge Regressor, and advanced gradient boosting decision tree (GBDT) algorithms such as XGBoost, LightGBM, and CatBoost, as well as three artificial neural networks (ANN). The predicted magnitude and risk zone of an earthquake are mapped using a geographic information system (GIS), and the maps performed well in terms of prediction. The generated maps is showing the expected earthquake risk based on historical data using the statistical computations. The ANN models perform exceptionally well, with R2 scores of 0.99 and 0.98 for training and case study data, respectively, and low values for MSE, MAE, and RMSE. ML models have demonstrated an exceptional ability to properly generalize from a single dataset, which implies they can accurately anticipates results for new and untested data. The results would be helpful to many local emergency preparedness and infrastructure planning organizations.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Overall, the results suggest that ANN-3 is a good model for the given dataset and can be used for further analysis or decisionmaking. Figure [36] displays the error function profile (Predicted value -Actual value) for ANN-3. This graph is useful in determining the profile of the error function.…”

Section: Evaluation Of Ann-3mentioning

confidence: 99%

Prediction of earthquake magnitude and seismic vulnerability mapping using artificial intelligence techniques: a case study of Turkey

Biswas

Kumar

Bera

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The human biological neural network inspires artificial neural networks (ANNs), and research on neural networks has been accompanied by an understanding and study of the human brain's structure and learning function [46]. There is no requirement for a set of special rules to solve the problem in this computational method, and the primary reliance is on the system's gradual training and learning [47]. Multi-layer perceptron networks (MLPs) are a type of artificial neural network algorithm widely used and successful for modeling and forecasting [48].…”

Section: Multi-layer Perceptron (Mlp)mentioning

confidence: 99%

A GIS-Based Assessment of Urban Tourism Potential with a Branding Approach Utilizing Hybrid Modeling

2022

View full text Add to dashboard Cite

Urban tourism has developed into one of the most efficient and effective economic and recreational activities in the modern era since the late second half of the twentieth century. Although attention to urban tourism development is increasing, this paper indicates that no comprehensive study has yet combined the two subjects of tourism development and branding for spatial modeling. Thus, for assessing urban tourism potential (AUTP), a novel hybrid modeling approach combining K-mean, fuzzy logic, and an artificial neural network (ANN) was used. The findings indicate that in Tabriz metropolis, areas 3 and 7 and the southwestern portion of region 6 have the most significant potential for urban tourism. In contrast, areas 4, 8, and 10 located within the worn-out urban fabric have the least potential for tourism. Subsequently, by examining the correlation between urban tourism conditioning factors (UTCFs) and tourism maps, it was determined that factors including distance from the catering centers, distance from the fault, quality of construction materials, distance from the historical centers, distance from the health centers, maximum temperature, and distance from the parking are the most critical in terms of increasing urban tourism's potential and branding. The analyses conducted in this study provide valuable and practical information for developing future strategies urban tourism. To this end, recommendations have been made to enhance tourism destination service delivery and management through branding.

show abstract

“…Adaptive resonance theory (ART)-based networks are a family of neural networks that have been used in spatial interaction flows, crop classification, and land-use change applications [134][135][136]. ART-based networks are supervised, self-organizing, and selfstabilizing neural networks that can learn fast in nonstationary environments [137].…”

Section: Adaptive Resonance Theory Networkmentioning

confidence: 99%

Machine Learning of Spatial Data

Nikparvar

Thill

2021

IJGI

View full text Add to dashboard Cite

Properties of spatially explicit data are often ignored or inadequately handled in machine learning for spatial domains of application. At the same time, resources that would identify these properties and investigate their influence and methods to handle them in machine learning applications are lagging behind. In this survey of the literature, we seek to identify and discuss spatial properties of data that influence the performance of machine learning. We review some of the best practices in handling such properties in spatial domains and discuss their advantages and disadvantages. We recognize two broad strands in this literature. In the first, the properties of spatial data are developed in the spatial observation matrix without amending the substance of the learning algorithm; in the other, spatial data properties are handled in the learning algorithm itself. While the latter have been far less explored, we argue that they offer the most promising prospects for the future of spatial machine learning.

show abstract

GIS-based seismic vulnerability mapping: a comparison of artificial neural networks hybrid models

Cited by 11 publications

References 71 publications

Prediction of earthquake magnitude and seismic vulnerability mapping using artificial intelligence techniques: a case study of Turkey

Prediction of earthquake magnitude and seismic vulnerability mapping using artificial intelligence techniques: a case study of Turkey

A GIS-Based Assessment of Urban Tourism Potential with a Branding Approach Utilizing Hybrid Modeling

Machine Learning of Spatial Data

Contact Info

Product

Resources

About