Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Recent developments in machine learning applications in landslide susceptibility

mapping
Na Kai Lun, Mohd Shahir Liew, Abdul Nasir Matori, and Noor Amila Wan Abdullah Zawawi

Citation: AIP Conference Proceedings 1905, 040022 (2017); doi: 10.1063/1.5012210


View online: https://doi.org/10.1063/1.5012210
View Table of Contents: http://aip.scitation.org/toc/apc/1905/1
Published by the American Institute of Physics

Articles you may be interested in


Heave motion prediction of a large barge in random seas by using artificial neural network
AIP Conference Proceedings 1905, 040017 (2017); 10.1063/1.5012205

Goal programming for cyclical auxiliary police scheduling at UiTM Cawangan Perlis
AIP Conference Proceedings 1905, 040021 (2017); 10.1063/1.5012209

Ontology for customer centric digital services and analytics


AIP Conference Proceedings 1905, 040024 (2017); 10.1063/1.5012212

Logistic regression and artificial neural network models for mapping of regional-scale landslide susceptibility in
volcanic mountains of West Java (Indonesia)
AIP Conference Proceedings 1730, 060001 (2016); 10.1063/1.4947407

Understanding the dynamic effects of returning patients toward emergency department density
AIP Conference Proceedings 1905, 040003 (2017); 10.1063/1.5012191

A modelling framework for mitigating customers’ waiting time at a vehicle inspection centre
AIP Conference Proceedings 1905, 040002 (2017); 10.1063/1.5012190
Recent Developments in Machine Learning Applications in
Landslide Susceptibility Mapping
Na Kai Lun 1, a), Mohd Shahir Liew 2, b) , Abdul Nasir Matori3, c) and Noor Amila
Wan Abdullah Zawawi4, d)
1,2,3,4
Civil & Environmental Engineering Department, Universiti Teknologi PETRONAS, Perak
2
Research & Innovation Office, Universiti Teknologi PETRONAS, Perak
a)
Corresponding author: nakailun@gmail.com
b)
shahir_liew@utp.edu.my
c)
nasrat@utp.edu.my
d)
amilawa@utp.edu.my

Abstract. While the prediction of spatial distribution of potential landslide occurrences is a primary interest in landslide
hazard mitigation, it remains a challenging task. To overcome the scarceness of complete, sufficiently detailed
geomorphological attributes and environmental conditions, various machine-learning techniques are increasingly applied
to effectively map landslide susceptibility for large regions. Nevertheless, limited review papers are devoted to this field,
particularly on the various domain specific applications of machine learning techniques. Available literature often report
relatively good predictive performance, however, papers discussing the limitations of each approaches are quite
uncommon. The foremost aim of this paper is to narrow these gaps in literature and to review up-to-date machine
learning and ensemble learning techniques applied in landslide susceptibility mapping. It provides new readers an
introductory understanding on the subject matter and researchers a contemporary review of machine learning
advancements alongside the future direction of these techniques in the landslide mitigation field.

BACKGROUND
Landslide susceptibility assessment (LSA) is an established approach as a preliminary solution to mitigating or
minimizing landslide damages [1]. It is a function of the likelihood of future landslide occurrences based on a
relative ranking of categories of each landslide causative parameter [2]. The resulting output is a solely spatial
distribution of the predicted categorized hazard probabilities across grid cells, known as landslide susceptibility
maps (LSM). LSA is generally categorized as qualitative and quantitative techniques in the evaluation. Pioneering
qualitative techniques are based upon expert knowledge and experience in defining the parameters involved and
susceptibility value in descriptive terms. Consequently, quantitative methods, such as deterministic and probabilistic
models, greatly diminish the subjectivity of heuristic (qualitative) techniques. While deterministic models based
upon physical properties of mass material are deemed the most effective approach, they are limited to local-scale
assessments with the availability of homogenous and high-resolution spatial features [3]. In practice, not all of
spatial layers can be economically and reliably determined over large regions to generate functional and accurate
maps [4], [5]. Consequently, probabilistic models are especially applicable for assessments over large regions
predisposed to landslide events with inadequate or inaccurate legacy and environmental information. The adeptness
of susceptibility methods is significantly dependent on the data distribution and size of datasets; therefore,
distribution-free multivariate techniques are ideal for landslide assessments [6]. The advent of soft computing has
enabled advanced quantitative approaches, such as machine learning and ensemble learning, to analyse extensive
spatial datasets made up of unstandardized data formats. It is widely acknowledged that there is no universally best
approach to landslide susceptibility mapping [7]. Each technique comes with its own distinctive challenges and
advantages. Surprisingly, only a handful of review papers are devoted to this field [8], [9], [10], [11] with a majority

Proceedings of the 13th IMT-GT International Conference on Mathematics, Statistics and their Applications (ICMSA2017)
AIP Conf. Proc. 1905, 040022-1–040022-6; https://doi.org/10.1063/1.5012210
Published by AIP Publishing. 978-0-7354-1595-9/$30.00

040022-1
published prior to 2013. The foremost aim of this paper is to narrow these gaps in literature, particularly for Asia,
where the threat of landslide mortality is high. The countries most affected are southwestern China, northern India,
southern Himalayas and northern Pakistan, Italy as well as regions in Central and South America [12].
This paper discusses more recent (2012-2017) publications on machine learning applications in LSA. More than
45 up-to-date and highly cited research papers from referred journals such as Environmental Earth Sciences, Natural
Hazards, Landslides, Remote Sensing, Geomorphology and International Journal of Remote Sensing have been
reviewed and a comparison on the different machine learning techniques used for regional scale hazard mapping is
presented.

RECENT DEVELOPMENTS OF MACHINE LEARNING APPLICATIONS IN LSM


Machine learning is a judicious fusion of applied mathematics and computational intelligence. It focuses on
‘training’ an algorithm to probe for and learn from data structure robust enough to make predictions; even without
predecessor knowledge of the structure. Machine learning algorithms are usually categorized according to the
approach of “learning”. The most common learning tasks are supervised learning techniques (for regression and
classification problems), such as logistic regression (LR) [13], [14], [15] Artificial Neural Network (ANN) [16], [17],
[18], [19] and Support Vector Machine (SVM) [20], [21] and unsupervised learning techniques (for clustering
purposes), such as k-means clustering and association rules. This paper also includes probabilistic machine learning
approaches for instance Evidential Belief Function (EBF), Certainty Factor (CF) and fuzzy logic applications. These
are of course not an exhaustive list of the techniques. Interested readers can refer to [22], [23] for further elucidation
on the concepts and differences. Available literature on machine learning applications in landslide susceptibility often
report relatively good predictive performance. There are also authors who found no difference in the predictive
performance of advanced techniques compared to conventional methods [24], [25].
Artificial Neural Network (ANN) is a popular method, which has robust tolerance towards noisy and incomplete
datasets and possesses rapid generalization capability [26], [27]. However, the disadvantage of ANN is its “black
box” model status, which prohibits straightforward assessment of the model’s validity. A hybrid neuro-fuzzy
weighting technique was applied in [28] to overcome the shortcomings of this black box nature of ANN. Fuzzy logic
is especially helpful when detailed data is not available. The authors derived weights for each category of landslide
causative factors using the Levenberg–Marquardt algorithm and ratings for each category using the cosine amplitude
method. The resulting susceptibility map was constructed using the weights and fuzzy set derived ratings. Relative to
the typical subjective weighting method, ANN method, and fuzzy set method, the hybrid technique constructed the
most accurate susceptibility map when measured up against existing field data [28].
Chauhan et al. [29] tested a modified ANN model, which allocated a weighted rating to every category of the
landslide causative factors. Forty-two categories, from seven causative factors, formed the input layer of the ANN.
Input neurons were allocated random discrete connection weights, which were continuously adjusted until a
satisfactory training accuracy was attained. The area under Receiver Operating Characteristic (ROC) curve
(AUROC) is a standard measure in landslide assessment to describe the predictive performance of a model. The
resulting modified ANN based map (AUROC = 0.88) performed better than the conventional ANN based map
(AUROC=0.84) in terms success rates and density analysis. Dou et al. [19] developed a hybrid certainty factor (CF)
and ANN model to predict landslide susceptibility in Japan using six predisposing causative features. The hybrid
model realized an AUROC of 0.82.
Evidential belief factor (EBF) is another machine learning algorithm which is gaining attention due to its
flexibility in uncertainty acceptance and its capacity to incorporate the probability confidence or belief of multiple
sources. Centred on the Dempster-Shafer theory, it is a generalization of the Bayesian theory of subjective
probability [30]. Using the same set of landslide causative factors, [30] compared the performance of an EBF model
and nine fuzzy logic models. Upon validation, all resulting models demonstrated high prediction capabilities with
area under curve (AUC) values ranging from 0.9185 to 0.9370, wherein the EBF model resulted in the best
performance. [31] established a runout propagation and landslide susceptibility map using data driven EBF. The
resulting map, built from nineteen landslide causative features, combined the landslide and spreading susceptibility
of the study site.
Jebur et al. [32] proved that the combination of EBF and RBF-SVM methods were highly effective for
susceptibility prediction compared to standalone methods of SVM and EBF. The radial basis function (RBF) kernel
was identified to work best for the selected case study. Moreover, SVM results in finer spatial structuring
particularly for spatially complex landscapes [33]. SVM is found to be advantageous in dealing with small training
sample numbers and a robust approach requiring minimum model tuning [34].

040022-2
Another study [35] for the same area tested two variations of neural networks namely the Bayesian Regularized
Neural Network (BRNN) and Levenberg–Marquardt Neural Network (LMNN). The study aims to explore the
variations to overcome the disadvantages of ANN being difficult to interpret at the level of individual features.
BRNN is an innovative approach while LMNN is an established robust method with good prediction capability due
to applied cluster analysis pre-processing and is deemed one of the faster ANNs approaches. The BRNN model was
independently trained while the LMNN model was divided into three subsets to prevent overfitting. The area under
the curve for BRNN and LMNN are 0.903 and 0.861 respectively, in comparison to a logistic regression model
(AUC= 0.938) implemented in an earlier study. BRNN not only outperformed LMNN, it was also a more robust
alternative for landslide susceptibility studies.
Bui et al. [36] studied the impacts of various kernel functions on SVM susceptibility models. The models were
SVM with linear function (LN-SVM), radial basis function (RBF-SVM), polynomial function (PL-SVM), and
sigmoid function (SIG-SVM). The grid search method with cross validation was applied to determine the “best”
classification parameters [37]. The same ten causative factors were employed and all maps were assessed using
prediction rate curves and Cohen’s Kappa index (Κ). The predictive capabilities of all resulting maps were
exceptional and comparable, with PL-SVM performing the best (AUC = 0.955, Κ=0.823), followed by RBF-SVM
(AUC=0.954, Κ=0.822), LN-SVM (AUC=0.952, Κ=0.756) and SIG-SVM (AUC=0.945,Κ=0.722). Κ values
represented the reliability of a model, the higher the value, the stronger the agreement between observed and
predicted values. These models performed better compared to logistic regression (AUC=0.938) and Bayesian
regularized neural network (AUC=0.903) models generated in earlier studies [36]. Bui et al. [38] compared the
performance of support vector machines (SVM) with two kernel functions, decision tree (DT) and Naïve Bayes
(NB) models in the computation of landslide susceptibility indexes for Hoa Binh in Vietnam. Subsequently, the
ensuing four susceptibility maps were contrasted against a map constructed based on logistic regression model by
the same researchers in 2011. The same ten parameters were used as the preceding study. A visual comparison of the
high susceptibility classes of the new maps demonstrates good agreement with field evidence and assumptions. The
predictive capability of the RBF-SVM and PL-SVM models were marginally superior to the logistic regression
model used in a previous study.
A comparison of five machine learning methods for a landslide susceptibility study in Uttarakhand concluded
SVM to be most feasible for the site compared to Logistic Regression (LR), Fisher's Linear Discriminant Analysis
(FLDA), Bayesian Network (BN), and Naïve Bayes (NB) [39]. The results were in line with prior case studies where
SVM was proven to outperform other learning methods such as logistic regression, Naïve Bayes and decision tree
[25], [33] , [40], [41]. The same authors further compared the performance of RBF-SVM and Naïve Bayes Trees
(NBT) for the spatial prediction of landslides for a mountainous Uttarakhand region. NBT is a hybrid of Naïve
Bayes and decision tree classifiers, which is an improvement of both techniques [42]. This study marked the first
application of NBT in landslide susceptibility assessment [43] and proved that NBT is a technique on par with SVM.
Bui et al. [44] evaluated the performance of five different landslide susceptibility model from support vector
machine (SVM), multi-layer perceptron neural networks (MLP Neural Nets), radial basis function neural networks
(RBF Neural Nets), kernel logistic regression (KLR), and logistic model trees (LMT). A list of twelve predisposing
causative factors was assessed beforehand with a ten-fold cross-validation technique. It was inferred that both KLR
and LMT performed the best for the training case study of landslides at Son La hydropower basin in Vietnam.
More recently, performance of Naïve Bayes (NB), Multilayer Perceptron (MLP) Neural Network and Functional
Tree (FT) were evaluated [45]. A part of the mountainous Uttarakhand region with coverage of about 323,815 m²
was selected as the study site. Naïve Bayes assumes that all features are provisionally independent to optimize
posterior probability in the classification process [46] and is ideal for small training datasets whilst retaining its
predictive performance [47]. These features are also a disadvantage as not all landslide features are independent of
one another and hinges on the absence of multicollinearity. In fact, the landslide causatives factors are usually
correlated [48]. MLP Neural Network is one of the most used ANN techniques [49]. It is robust and able to
generalize imprecise input data. However, users are required to define the optimal number of hidden layers first to
optimize errors in the training samples. Functional Tree is a hierarchical model to build multivariate trees, which
separates landslide and non-landslide classes using logistic regression function. This is the principal difference
between FT and the traditional decision tree models, where the latter divides input data at tree node with a constant
value [50]. For this study, the predictive capability of MLP Neural Network model was the best (AUC=0.851),
followed by FT model (AUC=0.849) and NB model (AUC=0.838) respectively. The authors hypothesized that the
excellent predictive performance of MLP was due to the optimization of input data weight values, not implemented
in FT and NB model [45].

040022-3
Pham et al. [51] further paired MLP Neural Network as a base classifier and various ensemble techniques to
evaluate the performance of each pair of ensemble frameworks. The study area covered 1325 km2 hilly environment
of northern part of Himalaya. Ensemble methods were first used to classify and optimize the input training data
before the base classifier groups the data into landslide susceptibility classes. The six ensemble techniques evaluated
were AdaBoost, Bagging, Dagging, MultiBoost, Rotation Forest and Random Subspace. The input data is made up
of fifteen landslide causative factors with each weighted using Relief-F method [52]. The higher the weight, the
more important the factor is for the framework. The input layer (landslide causative factors) was propagated forward
through a number of hidden layers to produce output values (inferring landslide or non-landslide classification). The
network was trained with 1 hidden layer and 500 epochs and the validation threshold is set to 20 upon a trial-and-
error course. As a standalone base classifier, the MLP neural net yielded an AUC value of 0.874. In comparison, the
ensemble framework with the MultiBoost model exhibited the best predictive capability (AUC=0.886) due to the
efficiency of MultiBoost in lowering both bias of the learning algorithm and variance of the learned model [53].

DISCUSSION & CONCLUSION


The use of hybrid machine learning techniques is observed to be picking up in contemporary literature. Applied
research as reviewed earlier has proven that these techniques are feasible solutions to the complex relationship
between input features. Ensemble learning combines multiple base machine learning classifiers to enhance their
performance by merging the benefits of both methods and accelerating processing time. These techniques are not
new and have been applied widely in the fields of computer science and banking. Nevertheless, the complexity of a
model is not the sole indicator to improved predictive performance. In designing an ideal landslide susceptibility
model, considerations such as input data quality and classification, model degree of freedom, reproducibility and
verification are crucial but, at times, overlooked in the process in proving “successful” models. To conclude,
although the predictive performances of most machine learning are comparable, each is unique in modeling the
relationships between the predictors and landslide initiation as well as ranking of relevant predictors, which is site-
specific. Insight into these differences is vital in the selection of a compatible technique for a particular study aim.
Attention should be emphasized in the data pre-processing, feature selection and model setup processes.

REFERENCES
1. R. Fell, J. Corominas, C. Bonnard, L. Cascini, E. Leroi and W. Savage, "Guidelines for landslide susceptibility,
hazard and risk zoning for land-use planning," Engineering Geology, vol. 102, no. 3, pp. 99-111, 2008.
2. D. Varnes, "Landslide Hazard Zonation: A Review of Principles and Practice. Review Report," Natural
Hazards, vol. 3, 1984.
3. C. van Westen and M. Terlien, "An approach towards deterministic landslide hazard analysis in GIS. A case
study from Manizales (Colombia)," Earth Surface Processes and Landforms, vol. 21, pp. 853-868, 1996.
4. C. van Westen, "Geo-information tools for landlside risk assessment - an overviw of recent developments," in
The 9th International Symposium on Landslides, Rio de Janeiro, 2004.
5. J. Graff, H. Romesburg, R. Ahmad and J. McCalpain, "Producing landslide-susceptibility maps for regional
planning in data-scarce regions," Natural Hazards, vol. 64, pp. 729-749, 2012.
6. E. Karimi Sangchini, M. Ownegh, A. Sadoddin and A. Mashayekhan, "Probabilistic landslide risk analysis and
mapping (Case Study: Chehel-Chai watershed, Golestan Province, Iran)," Journal of Rangeland Science, vol. 2,
no. 1, pp. 425-436, 2011.
7. J. Goetz, A. Brenning, H. Petschko and P. Leopold, "Evaluating machine learning and statistical prediction
techniques for landslide susceptibility modeling," Computers and Geosciences, vol. 81, pp. 1-11, 2015.
8. P. Aleotti and R. Chowdhury, "Landslide hazard assessment: summary review and new perspectives," Bulletin
of Engineering Geology and the Environment, vol. 58, pp. 21-44, 1999.
9. C. Baeza and J. Corominas, "Assessment of shallow landslide susceptibility by means of multivariate statistical
techniques," Earth Surface Processes and Landforms, vol. 26, pp. 1251-1263, 2001.
10. S. Pardeshi, S. Autade and S. Pardeshi, "Landslide hazard assessment: recent trends and techniques,"
Springerplus, vol. 2, p. 523, 2013.
11. B. Pradhan, M. Seeni and B. Kalantar, "Performance Evaluation and Sensitivity Analysis of Expert-Based,
Statistical, Machine Learning, and Hybrid Models for Producing Landslide Susceptibility Maps," in Laser
Scanning Applications in Landslide Assessment, B. Pradhan, Ed., Springer, 2017, pp. 193-232.

040022-4
12. W. Yang, L. Shen and P. Shi, "Mapping Landslide Risk of the World," in World Atlas of Natural Disaster
Risk, P. Shi and R. Kaperson, Eds., Springer and BNUP, 2015, pp. 57-66.
13. L. Ayalew and H. Yamagishi, "The application of GIS-based logistic regression for landslide susceptibility
mapping in the Kakuda-Yahiko Mountains, Central Japan," Geomorphology, vol. 65, no. 1-2, pp. 15-31, 2005.
14. S. Lee, "Application of logistic regression model and its validation for landslide susceptibility mapping using
GIS and remote sensing data," International Journal of Remote Sensing, vol. 26, no. 7, pp. 1477-1491, 2005.
15. A. Akgun, "A comparison of landslide susceptibility maps produced by logistic regression, multi-criteria
decision, and likelihood ratio methods: a case study at İzmir, Turkey," Landslides, vol. 9, no. 1, pp. 93-106,
2012.
16. S. Pascale, S. Parisi, A. Mancini, M. Schiattarella, M. Conforti, A. Sole, B. Murgante and F. Sdao, "Landslide
Susceptibility Mapping Using Artificial Neural Network in the Urban Area of Senise and San Costantino," in
International Conference on Computational Science and Its Applications, Basilicata, Southern Italy, 2013.
17. B. Pradhan and S. Lee, "Delineation of landslide hazard areas on Penang Island, Malaysia, by using frequency
ratio, logistic regression, and artificial neural network models," Environmental Earth Sciences, vol. 60, no. 5,
pp. 1037-1054, 2010.
18. P. Tsangaratos and A. Benardos, "Estimating landslide susceptibility through a artificial neural network
classifier," Natural Hazards, vol. 74, pp. 1489-1516, 2014.
19. J. Dou, H. Yamagishi, H. Pourghasemi, A. Yunus, X. Song, Y. Xu and Z. Zhu, "An integrated artificial neural
network model for the landslide susceptibility assessment of Osado Island, Japan," Natural Hazards, vol. 78,
pp. 1749-1776, 2015.
20. O. Mora, J. Liu, M. Lenzano, C. Toth and D. Gejner-Brzezinska, "Small landslide susceptibility and hazard
assessment based on airborne lidar data," Photogrammetric Engineering & Remote Sensing, vol. 81, no. 3, pp.
239-247, 2015.
21. R. Mahalingam and M. Olsen, "Evaluation of the influence of source and spatial resolution of DEMs on
derivative products used in landslide mapping," Geomatics, Natural Hazards and Risk, vol. 7, no. 6, pp. 1835-
1855, 2016.
22. I. Witten, E. Frank, M. Hall and C. Pal, Data Mining - Practical Machine Learning Tools and Techniques 4th
Edition, San Francisco, CA: Elsevier, 2016.
23. T. Hastie, R. Tibshirani and J. Friedman, The Elements of Statistical Learning : Data Mining, Inference and
Prediction Second Edition, New York: Springer-Verlag, 2009.
24. C. Poudyal, C. Chang, H. Oh and S. Lee, "Landslide susceptibility maps comparing frequency ratio and
artificial neural networks: a case study from the Nepal Himalaya," Environmental Earth Sciences, vol. 61, no.
5, pp. 1049-1064, 2010.
25. T. Kavzoglu, E. Sahin and I. Colkesen, "Landslide susceptibility mapping using GIS-based multi-criteria
decision analysis, support vector machines and logistic regression," Landslides, vol. 11, pp. 425-439, 2014.
26. H. Shi, K. Lee, H. Lee , W. Ho, D. Sun, J. Wang and C. Chiu, "Comparison of artificial neural network and
logistic regression models for predicting in-hospital mortality after primary liver cancer surgery," PloS one,
vol. 7, no. 4, 2012.
27. H. Yuan, C. Van Der Wiele and S. Khorram, ". An automated artificial neural network system for land
use/land cover classification from Landsat TM imagery," Remote Sensing, vol. 9, p. 498, 2009.
28. D. Kanungo, M. Arora, S. Sarkar and R. Gupta, "A comparative study of conventional, ANN black box, fuzzy
and combined neural and fuzzy weighting procedures for landslide susceptibility zonation in Darjeeling
Himalayas," Engineering Geology, vol. 85, no. 3-4, pp. 347-366, 2006.
29. S. Chauhan, M. Sharma, M. Arora and N. Gupta, "Landslide susceptibility zonation through ratings derived
from artificial neural network," International Journal of Applied Earth Observation and Geoinformation, vol.
12, pp. 340-350, 2010.
30. D. Bui, B. Pradhan, O. Lofman, I. Revhaug and O. Dick , "Spatial prediction of landslide hazards in Hoa Binh
province (Vietnam): A comparative assessment of the efficacy of evidential belief functions and fuzzy logic
models," Catena, vol. 96, pp. 28-40, 2012a.
31. A. Pradhan, H. Kang, S. Lee and Y. Kim, "Spatial model integration for shallow landslide susceptibility and its
runout using a GIS-based approach in Yongin, Korea," Geocarto International, vol. 32, no. 4, pp. 420-441,
2017.
32. M. Jebur, B. Pradhan and M. Tehrany, "Manifestation of LiDAR-derived parameters in the spatial prediction
of landslides using novel ensemble evidential belief functions and support vector machine models in GIS,"

040022-5
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 8, no. 2, pp. 674-
690, 2015.
33. A. Brenning, "Spatial prediction models for landslide hazards: review, comparison and evaluation," Natural
Hazards and Earth System Sciences, vol. 5, pp. 853-862, 2005.
34. C. Huang, L. Davis and J. Townshend, "An assessment of support vector machines for land cover
classification," International Journal of Remote Sensing, vol. 23, pp. 725-749, 2002.
35. D. Bui, B. Pradhan, O. Lofman, I. Revhaug and O. Dick, "Landslide susceptibility assessment in the Hoa Binh
province of Vietnam : A comparison of the Levenberg-Marquardt and Bayesian regularized neural networks,"
Geomorphology, vol. 171, pp. 12-29, 2012b.
36. D. Bui, B. Pradhan, O. Lofman, I. Revhaug and O. Dick, "Application of support vector machines in landslide
susceptibility assessment for the Hoa Binh province (Vietnam) with kernel functions analysis," in International
Congress on Environmental Modelling and Software, Leipzig, Germany, 2012c.
37. L. Zhuang and H. Dai, "Parameter optimization of kernel-based one-class classifier on imbalance text
learning," Journal of Computers, vol. 1, no. 7, pp. 32-40, 2006.
38. D. Bui , B. Pradhan, O. Lofman and I. Revhaug, "Landslide susceptibility assessment in Vietnam using support
vector machines, decision tree, and Naıve Bayes Models," Mathematical Problems in Engineering, 2012d.
39. B. Pham, B. Pradhan, D. Bui, I. Prakash and M. Dholakia, "A comparative study of different machine learning
methods for landslide susceptibility assessment : A case study of Uttarakhand area (India)," Environmental
Modelling and Software, vol. 84, pp. 240-250, 2016a.
40. M. Marjanovic, M. Kovacevic, B. Bajat and V. Vozenilek, "Landslide susceptibility assessment using SVM
machine learning algorithm," Engineering Geology, vol. 123, no. 3, pp. 225-234, 2011.
41. X. Yao, L. Tham and F. Dai, "Landslide susceptibility mapping based on Support Vector Machine: a case
study on natural slopes of Hong Kong, China," Geomorphology, vol. 101, no. 4, pp. 572-582, 2008.
42. R. Kohavi, "Scaling Up the Accuracy of Naive Bayes Classifiers: A Decision-Tree Hybrid," in Knowledge
Discovery and Data Mining, Portland, Oregon, 1996.
43. B. Pham, D. Bui, I. Prakash and M. Dholakia, "Evaluation of predictive ability of support vector machines and
naive Bayes trees methods for spatial prediction of landslides in Uttarakhand state (India) using GIS," Journal
of Geomatics, vol. 10, no. 1, pp. 71-79, 2016b.
44. D. Bui, T. Tuan, H. Klempe, B. Pradhan and I. Revhaug, "Spatial prediction models for shallow landslide
hazards: a comparative assessment of the efficacy of support vector machines, artificial neural networks, kernel
logistic regression, and logistic model tree," Landslides, vol. 13, no. 2, pp. 361-378, 2016.
45. B. Pham, D. Bui, H. Pourghasemi, P. Indra and M. Dholakia, "Landslide susceptibility assesssment in the
Uttarakhand area (India) using GIS: a comparison study of prediction capability of naïve bayes, multilayer
perceptron neural networks, and functional trees methods," Theoretical and Applied Climatology, vol. 128, no.
1-2, pp. 255-273, 2015.
46. J. Soni, U. Ansari, D. Sharma and S. Soni, "Predictive data mining for medical diagnosis: an overview of heart
disease prediction," International Journal of Computer Applications, vol. 17, pp. 43-48, 2011.
47. P. Tsangaratos and I. Ilia, "Comparison of logistic regression and Naive Bayes classifier in landslide
susceptibility assessments: The influence of models complexity and training dataset size," Catena, vol. 145, pp.
164-179, 2016.
48. D. Bui, T. Tuan, H. Klempe, B. Pradhan and I. Revhaug, "Spatial prediction models for shallow landslide
hazards: a comparative assessment of the efficacy of support vector machines,artificial neural networks, kernel
logistic regression, and logistic model tree," Landslides, vol. 13, no. 2, pp. 1-18, 2015.
49. M. Zare, H. Pourghasemi, M. Vafakhah and B. Pradhan, "Landslide susceptibility mapping at Vaz Watershed
(Iran) using an artificial neural network model: a comparison between multilayer perceptron (MLP) and radial
basic function (RBF) algorithms," Arabian Journal of Geosciences, vol. 6, no. 8, pp. 2873-2888, 2013.
50. H. Lan , E. Frank and M. Hall, Data mining : Practical machine learning tools and techniques, Boston: Morgan
Kaufman, 2011.
51. B. Pham, D. Bui, I. Prakash and M. Dholakia, "Hybrid integration of Multilayer Perceptron Neural Networks
and machine learning ensembles for landslide susceptibility assessment at Himalayan area (India) using GIS,"
CATENA, vol. 149, pp. 52-63, 2017.
52. P. Langley, "Selection of relevant features in machine learning," in 1994 Fall Symposium, Menlo Park,
California, 1994.
53. S. Joshi and V. Pimprale, "Network Intrusion Detection System (NIDS) based on Data Mining," International
Journal of Engineering Science and Innovative Technology (IJESIT), vol. 2, no. 1, pp. 95-98, 2013.

040022-6

You might also like