CN113822522A - Landslide susceptibility assessment method, device and equipment and readable storage medium - Google Patents
Landslide susceptibility assessment method, device and equipment and readable storage medium Download PDFInfo
- Publication number
- CN113822522A CN113822522A CN202110698085.5A CN202110698085A CN113822522A CN 113822522 A CN113822522 A CN 113822522A CN 202110698085 A CN202110698085 A CN 202110698085A CN 113822522 A CN113822522 A CN 113822522A
- Authority
- CN
- China
- Prior art keywords
- landslide
- disaster
- susceptibility
- landslide susceptibility
- causing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000005457 optimization Methods 0.000 claims abstract description 42
- 238000012216 screening Methods 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000013135 deep learning Methods 0.000 claims abstract description 19
- 238000011156 evaluation Methods 0.000 claims description 44
- 238000007637 random forest analysis Methods 0.000 claims description 20
- 238000011160 research Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 239000002689 soil Substances 0.000 claims description 6
- 230000000007 visual effect Effects 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 5
- 238000011835 investigation Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims description 2
- 230000002265 prevention Effects 0.000 abstract description 7
- 238000012706 support-vector machine Methods 0.000 description 11
- 230000009467 reduction Effects 0.000 description 9
- 238000013136 deep learning model Methods 0.000 description 8
- 238000010801 machine learning Methods 0.000 description 6
- 229910052582 BN Inorganic materials 0.000 description 5
- PZNSFCLAULLKQX-UHFFFAOYSA-N Boron nitride Chemical compound N#B PZNSFCLAULLKQX-UHFFFAOYSA-N 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 238000010276 construction Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000013210 evaluation model Methods 0.000 description 4
- 230000001939 inductive effect Effects 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000005065 mining Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 239000003016 pheromone Substances 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- 241000257303 Hymenoptera Species 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009024 positive feedback mechanism Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
Abstract
The invention provides a landslide susceptibility assessment method, a landslide susceptibility assessment device, equipment and a readable storage medium, and relates to the technical field of geological disaster prevention and control, wherein the method comprises the following steps: screening out landslide susceptibility disaster-causing factors in the sample area; combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set; the method comprises the steps of taking the disaster-causing factor-sample set as input data for training, and obtaining an ant colony optimization algorithm-deep belief network model for generating a landslide susceptibility result of a region to be evaluated in a deep learning mode.
Description
Technical Field
The invention relates to the technical field of geological disaster prevention and control, in particular to a landslide susceptibility assessment method, a landslide susceptibility assessment device, equipment and a readable storage medium.
Background
Landslide susceptibility refers to the possibility of landslide occurrence evaluation by comprehensively analyzing geological and environmental conditions in a research area, landslide susceptibility evaluation research is the primary link of geological disaster risk evaluation, and achievements of landslide susceptibility evaluation research can provide application services for scientifically and economically organizing and implementing disaster prevention and reduction measures,
the method can also provide scientific and technological support for the formulation of emergency schemes in areas threatened by geological disasters, also provide decision reference for the suitability of site selection of construction projects such as infrastructure and the like and the rationality evaluation of spatial layout of the construction projects, and provide scientific basis for medium-term and long-term construction planning of the areas.
The landslide susceptibility evaluation model is an important pivot for revealing potential complex relation between landslide susceptibility disaster-causing factors and samples, and the prediction capability of the model is also determined by considering the applicability of the model according to the characteristics of specific areas, so that the selection of the landslide susceptibility evaluation model is an important link in a landslide susceptibility evaluation flow.
In the existing model method, a physical landslide susceptibility assessment model requires a large amount of detailed data, however, data collection is difficult and laborious for extensive monitoring. A heuristic driving model is constructed based on limited information, and slope disaster-causing factors are ranked or weighted by expert opinions and professional knowledge to be parameterized, so that the method is difficult to objectively quantify or evaluate results. For a traditional shallow machine learning model, the traditional shallow machine learning model has a shallow structure with only one or zero hidden layers, and has the defects of limited training time, easy falling into local optimum, unstable convergence and the like.
Therefore, a landslide susceptibility assessment method for mining deeper links between data to effectively solve the problems of overfitting, difficult convergence, too slow and the like is an important issue to be solved in the industry at present.
Disclosure of Invention
The invention provides a landslide susceptibility assessment method, a landslide susceptibility assessment device, equipment and a readable storage medium, which are used for solving the defects of overfitting, difficult convergence, too slow and the like of the landslide susceptibility assessment method in the prior art and realizing more values of a deep learning model in disaster prevention and reduction applications such as landslide susceptibility and the like.
The invention provides a landslide susceptibility assessment method, which comprises the following steps:
screening out landslide susceptibility disaster-causing factors in the sample area;
combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
and taking the disaster-causing factor-sample set as input data for training, and obtaining an ant colony optimization algorithm-deep belief network model for generating a landslide proneness result of the area to be evaluated in a deep learning mode.
According to the landslide susceptibility assessment method provided by the invention, the landslide susceptibility disaster-causing factor is screened out, and the method specifically comprises the following steps:
collecting landslide information of a sample area; the landslide information comprises remote sensing images, satellite images, basic geography, geology, landform, meteorological hydrology, forest land types, soil types, historical landslide disaster records and landslide research literature data;
and screening the landslide susceptibility disaster-causing factors corresponding to the landslide information from the landslide information by using the Pearson correlation coefficient and the importance ranking of random forest features.
According to the landslide susceptibility assessment method provided by the invention, the landslide sample set collected from the sample area specifically comprises the following steps:
a landslide sample set of sample areas is collected using regional landslide investigation, visual interpretation based on satellite images, and intelligent identification based on remote sensing imagery.
According to the landslide susceptibility evaluation method provided by the invention, area landslide investigation, visual interpretation based on satellite images and intelligent identification based on remote sensing images are utilized to acquire landslide samples collected by sample areas and convert area data into point data.
According to the landslide susceptibility evaluation method provided by the invention, in the training process of the ant colony optimization algorithm-depth belief network model, the heuristic ant colony algorithm is used for searching the optimal path and continuously iterating to search the optimal combination and optimal parameters of the ant colony optimization algorithm-depth belief network model until the iteration times are maximum, and the ant colony optimization algorithm-depth belief network model with the optimal combination and parameters is obtained.
According to the landslide susceptibility evaluation method provided by the invention, in the parameters of an ant colony optimization algorithm-deep belief network model, an activation function adopts a hyperbolic tangent function, an optimizer adopts an adaptive moment estimation optimizer, an optimization strategy adopts a batch standardization optimization strategy, a ridge regression weight attenuation item regularization and early stop method is used, and a loss function adopts a cross entropy and a ridge regression adjustment item.
According to the landslide susceptibility evaluation method provided by the invention, the structure of the ant colony optimization algorithm-depth belief network model comprises an input layer, an output layer and four layers of limited Boltzmann machine layers positioned between the input layer and the output layer, wherein the number of neurons in the four layers of limited Boltzmann machine layers is gradually reduced from the input layer to the output layer.
The invention also provides a landslide susceptibility assessment device, comprising:
the disaster-causing factor screening module is used for screening out landslide susceptibility disaster-causing factors in the sample area;
the sample set establishing module is used for combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
and the proneness evaluation module is used for taking the disaster-causing factor-sample set as input data used for training and obtaining an ant colony optimization algorithm-deep belief network model for generating a landslide proneness result of the area to be evaluated in a deep learning mode.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of the landslide susceptibility assessment method.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the landslide susceptibility assessment method as any one of above.
According to the landslide susceptibility evaluation method, the device, the equipment and the readable storage medium, the deep belief network model and the ant colony optimization algorithm are combined and applied to landslide susceptibility evaluation, the problems of model overfitting, slow convergence and the like are solved by optimizing combination and parameters in an optimization strategy of the deep belief network model through the ant colony optimization algorithm, the landslide susceptibility disaster-causing factors of the sample area are screened out, and the landslide susceptibility disaster-causing factors with weak correlation characteristics are preferably selected out, so that deeper relation between the landslide susceptibility disaster-causing factors and the sample is realized, and finally, more values of the deep learning model in disaster prevention and reduction applications such as landslide susceptibility and the like are realized.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a landslide susceptibility assessment method provided by the present invention;
FIG. 2 is a schematic flow chart of the landslide susceptibility assessment method provided by the invention when constructing an ACO-DBN model;
FIG. 3 is a graph of the variation trend of the time training set and the validation set accuracy and loss along with the number of iteration rounds when the landslide susceptibility assessment method provided by the present invention constructs an ACO-DBN model;
FIG. 4 is a chart result of susceptibility of the Jiuzhaigou landslide area drawn in the landslide susceptibility evaluation method provided by the present invention;
FIG. 5 is a structural diagram of an ACO-DBN model in the landslide susceptibility assessment method provided by the present invention;
FIG. 6 is a graph of ROC curves and corresponding AUC values of four models in a verification stage by the landslide susceptibility assessment method provided by the invention;
fig. 7 is a specific flowchart of step S100 in the landslide susceptibility assessment method provided by the present invention;
FIG. 8 is a schematic structural diagram of a landslide susceptibility assessment apparatus provided by the present invention;
FIG. 9 is a schematic structural diagram of a disaster-causing factor screening module in the landslide susceptibility assessment apparatus according to the present invention
Fig. 10 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The landslide susceptibility evaluation model is an important pivot for revealing potential complex relation between landslide susceptibility disaster-causing factors and samples, and the prediction capability of the model is also determined by considering the applicability of the model according to the characteristics of specific areas, so that the selection of the landslide susceptibility evaluation model is an important link in a landslide susceptibility evaluation flow. With the successful application of more and more landslide susceptibility assessment algorithms, models and methods in landslide susceptibility research, at present, the landslide susceptibility assessment algorithms, models and methods are classified into two categories of determinacy and non-determinacy. The model belonging to the deterministic category has the advantages of definite physical significance, high prediction precision, timely feedback and the like, but is difficult to acquire or has high cost for large-range monitoring data. The other type of the method is based on data driving and can be subdivided into statistical models and Machine learning models, such as Frequency Ratio models (FR), Support Vector Machines (SVM), Random Forest (RF), Back Propagation Neural Networks (BPNN), and the like, the statistical modeling reveals the relationship between variables in data through mathematical equations, the Machine learning learns independent rule functions through data, and the logistic regression and Bootstrap refer to the statistical models into Machine learning models.
In the existing model method, a physical landslide susceptibility assessment model requires a large amount of detailed data, however, data collection is difficult and laborious for extensive monitoring. A heuristic driving model is constructed based on limited information, and slope disaster-causing factors are ranked or weighted by expert opinions and professional knowledge to be parameterized, so that the method is difficult to objectively quantify or evaluate results. For the traditional shallow machine learning model, such as a shallow structure with only one or zero hidden layers, such as an SVM and an RF, the method has the defects of limited training time, easy falling into local optimum, unstable convergence and the like. With the continuous emergence of deep learning results, a new thought and method are provided for landslide proneness assessment research. Deep learning learns more complex, higher-level hidden features by analyzing the features hierarchically. Therefore, aiming at the complex nonlinear relationship between the disaster-causing factors and the samples in the landslide susceptibility research, the deep learning method is helpful for mining deeper links between data, and effectively solves the problems of overfitting, difficult convergence, too slow speed and the like through a series of optimization strategies such as activating partial neurons (Dropout), Batch standardization (BN), Adaptive moment estimation (Adam) optimizer, Early Stopping method (Early Stopping) and the like. Although deep learning research is carried out in the related field, few researches are carried out on disaster-causing factor applicability, deep learning optimization strategies and multi-parameter optimization aiming at different geographic environments, so that the existing deep learning-based landslide vulnerability assessment method still needs to be improved.
By combining the above, a landslide susceptibility assessment method for mining deeper links between data to effectively solve the problems of overfitting, difficult convergence, too slow and the like is an important issue to be solved in the industry at present.
The landslide susceptibility assessment method of the present invention is described below with reference to fig. 1, comprising the steps of:
s100, screening out landslide susceptibility disaster-causing factors in the sample area. Due to the fact that the selection of the landslide susceptibility disaster-causing factors lacks theoretical basis and the landslide susceptibility disaster-causing factors possibly have strong correlation under different geographic environments, meanwhile, in a deep learning model, due to the fact that too large feature dimensionality easily causes reduction of model overfitting and execution efficiency, abnormal noise features of too many samples are learned, and the generalization capability of the model is also reduced, and therefore the landslide susceptibility disaster-causing factors need to be correspondingly screened. After the processing of step S100, the obtained landslide susceptibility disaster-causing factors are all weak-related geographical-environmental factors, i.e., weak-related features, which can exclude strong-related features, and at the same time, exclude features that contribute little to model training, so that the efficiency and accuracy of the subsequently generated model are improved.
S200, combining the landslide susceptibility disaster-causing factors with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set.
S300, taking the disaster-causing factor-sample set as input data used for training, and obtaining an ant colony optimization algorithm-deep belief network model for generating a landslide proneness result of the area to be evaluated in a deep learning mode.
At present, although deep learning research is carried out in the field of landslide susceptibility assessment, few researches are carried out on the applicability of disaster-causing factors in different geographic environments, deep learning optimization strategies and multi-parameter optimization. Meanwhile, deep learning has problems such as structure of the model, generalization and robustness of the model, difficulty in overfitting and convergence and the like, and optimization of the model in landslide susceptibility assessment research is not considered yet.
In step S300, an Ant Colony Optimization (ACO) -Deep Belief Network (DBN) model is adopted, and the DBN model is applied to landslide susceptibility assessment by combining with ACO for the first time in the landslide susceptibility assessment method of the present invention.
Aiming at the parameters of a deep belief network and an ACO-DBN model, an activation function adopts a hyperbolic tangent (tanh) function, an optimizer adopts an Adam optimizer, an optimization strategy adopts a BN (boron nitride) optimization strategy, an L2 weight attenuation term regularization, an early stop method, a Dropout and the like are used, a loss function adopts cross entropy and an L2 adjustment term, and the Dropout and the BN parameters are optimized simultaneously through an ant colony algorithm, so that higher precision is ensured on the basis of solving the problems of overfitting and too slow convergence. The ACO is an evolutionary algorithm based on population recently proposed, a heuristic population searching intelligent algorithm based on population optimization, wherein positive feedback mechanisms are formed by pheromone communication between ants in an iterative search process, and the ACO has strong robustness and global search capability in solving performance, is essentially parallel and is insensitive to initial value selection. Accumulating pheromones as the core of ant colony algorithm updating, and the calculation formula of ACO is as follows:
Tij(t+1)＝(1-ρ)·τij(t)+Δτ
wherein tau ij represents the pheromone concentration between the node i and the node j at the time t, rho is the probability of searching k times of ants moving from the node i to the node j, and the ant period model is selected by updating delta tau.
Referring to fig. 2, in the training process of the ACO-DBN model, a heuristic ant colony algorithm is used to find the optimal path and iterate continuously to find the optimal combination of the ACO-DBN model and the optimal parameters, including the learning rate and the minimum batch size in the BN, until the iteration number is maximum. And selecting an optimal parameter combination, and finally converging the ACO-DBN model through about 180 iteration rounds. Referring to fig. 3, during the training process of the ACO-DBN model, the difference between the accuracy of the training set and the accuracy of the validation set is kept at about 0.02, which indicates that the ACO-DBN model has no obvious overfitting phenomenon. Please refer to fig. 4, which illustrates an example of applying the method to landslide caused by an earthquake in nine villages, and performing landslide susceptibility evaluation, fig. 4 is a result of mapping landslide susceptibility in an earthquake region in nine villages, where the iteration number is the largest, and an ACO-DBN model with an optimal combination and parameter is obtained, and a corresponding susceptibility mapping is performed by using the combination and the ACO-DBN model with the optimal parameter.
Referring to fig. 5, the architecture of the ACO-DBN model includes an input layer, an output layer, and a four-layer Restricted Boltzmann Machine (RBM) layer located between the input layer and the output layer, wherein the number of neurons in the four-layer RBM layer decreases from the input layer to the output layer. The output layer is also called a softmax regression layer, and softmax regression is a generalization of Logistic regression to multi-class problems and is generally used for the output layer of a neural network.
In this embodiment, the connection between the layers of the six-layer framework is 9-400-200-50-2, where 9 is the number of disaster-causing factors prone to landslide, 400, 200, 100, and 50 are the numbers of neurons corresponding to the four layers of RBMs, respectively, and 2 is the probability of landslide and non-landslide prediction through the Softmax regression layer.
According to the landslide susceptibility evaluation method provided by the invention, the DBN model and the ACO are combined and applied to landslide susceptibility evaluation, the problems of model overfitting, slow convergence and the like are solved by optimizing the combination and parameters in the optimization strategy of the DBN model through the ACO, the landslide susceptibility disaster-causing factors of the sample area are screened out, and the landslide susceptibility disaster-causing factors with weak correlation characteristics are optimized, so that deeper relation between the landslide susceptibility disaster-causing factors and the sample is realized, and finally, more values of the deep learning model in landslide susceptibility and other disaster prevention and reduction applications are realized.
In the following, the landslide susceptibility assessment method of the present invention is described with reference to fig. 7, and step S100 specifically includes the following steps:
and S110, collecting landslide information of the sample area. The landslide information comprises remote sensing images, satellite images, basic geography, geology, terrain and geomorphology, meteorological hydrology, forest land types, soil types, historical landslide disaster records and landslide research literature data, and the remote sensing images, the satellite images, the basic geography, geology, the terrain and geomorphology, the meteorological hydrology, the forest land types and the soil types are basic data.
S120, screening out a landslide susceptibility disaster-causing factor corresponding to the landslide information from the landslide information by using the Pearson correlation coefficient and the random forest feature importance ranking.
In step S120, for different geographic environments, the selection of the landslide susceptibility disaster-causing factor lacks a theoretical basis, and the disaster-causing factor may have strong correlation, and meanwhile, in the deep learning model, too large feature dimension may cause reduction in model overfitting and execution efficiency, and abnormal noise features of too many samples may also reduce generalization capability of the model, so that the landslide susceptibility disaster-causing factor needs to be screened. And screening the disaster-causing factors through the Pearson correlation coefficient and the importance ranking of the random forest features so as to improve the operation efficiency of the model and ensure the accuracy. The pearson correlation coefficient characterizes the degree of correlation between two evaluation factors by a specific value between-1 and 1. When the two evaluation factors are independent from each other and have no correlation, the corresponding correlation coefficient value is 0; when the two evaluation factors accord with the linear positive correlation, the correlation coefficient value between the two evaluation factors is 1; when the correlation coefficient values of the two evaluation factors are-1, it is indicated that a linear negative correlation relationship is satisfied between the factors. The random forest feature importance ordering is to directly measure the influence of each feature on the model prediction accuracy according to random forests, and the basic idea is to rearrange the sequence of a certain list of feature values and observe the accuracy of the model reduced. By the two methods, not only can the strongly relevant characteristics be eliminated, but also the obtained landslide susceptibility disaster-causing factors are the weakly relevant geographical-environmental factors, namely the weakly relevant characteristics, and the characteristics with small contribution to model training are eliminated, so that the efficiency and the precision of the ACO-DBN model are improved.
After the processing in steps S110 and S120, finally, in the landslide susceptibility evaluation method of the present invention, 9 screened landslide susceptibility disaster-inducing factors are respectively lithology, slope, distance from road, Normalized Difference Vegetation Index (NDVI), slope position, distance from water system, land utilization, curvature, and slope direction, and it can be understood that the above 9 landslide susceptibility disaster-inducing factors participate in the construction process of the ACO-DBN model.
Step S200 specifically includes the following steps:
the landslide sample set of the sample area is collected by utilizing regional landslide investigation, visual interpretation based on satellite images and intelligent identification based on remote sensing images, and the three methods can complement and verify each other.
In this embodiment, in step S200, the partial surface landslide data is also converted into uniform point data (elements) by using Geographic Information System (Geo-Information System, GIS) software or the like.
The method is explained by taking the application to landslide caused by the earthquake of Jiuzhaigou as an example for landslide susceptibility evaluation, the ACO-DBN model constructed by the landslide susceptibility evaluation method finally selects 7184 sample points with the positive and negative sample ratio of 1:1, wherein 60% of sample data is used as a training set, and 40% of sample data is used as a test set
In the verification stage of the performance of the ACO-DBN model, an RF model, an SVM model, a DBN model and the ACO-DBN model are used for precision evaluation, wherein the RF model and the SVM model are based on RapidMiner Studio when a landslide susceptibility model is built, and a data structure, a modeling flow and the DBN model are kept consistent so as to be beneficial to comparative analysis. By adopting an accuracy priority criterion, parameters adopt an automatic optimization strategy, and finally through training and adjustment, main parameters of the RF are selected as follows: the number of decision trees is 9, the maximum depth is 4, the size of the minimum leaf node is 3, and the minimum gain coefficient is 0.01; the SVM kernel function selects a radial basis, the gamma value of the radial basis function is 0.1, and the penalty factor C is 1.2.
In order to verify the performance and stability of the ACO-DBN model, 2874 test samples (the proportion of positive samples to negative samples is 1:1) are adopted to carry out precision evaluation on the four models in a verification stage, 5 single-threshold statistical indexes of accuracy, precision, specificity and sensitivity and an F1 value (F1 Score) and multiple-threshold indexes of ROC curves and AUC values are selected to comprehensively evaluate the prediction capability of the models, namely the precision of the models, and the prediction capabilities of the four models are shown in Table 1.
Table 1 comparison table of prediction ability of four models in verification stage
Referring to table 1 and fig. 6, the accuracy of the ACO-DBN model test is 92.97%, the next time the DBN and SVM models are tested is the lowest RF, which indicates that the overall performance of the ACO-DBN model classifier is better. The accuracy rate reflects the probability that the landslide unit is correctly classified as a positive sample, namely, the landslide unit is predicted, in the landslide susceptibility analysis, more landslide units are expected to be correctly predicted, the accuracy rate of the RF model is higher than that of the SVM model, the accuracy rate is inconsistent with that of the SVM model, and the model is difficult to optimize; and the ACO-DBN model is superior to the comparison model in both aspects, and shows better consistency. On the other hand, the specificity and sensitivity indexes reflect the probability of being correctly predicted in real positive and negative samples, so that the performance of the SVM model for predicting the positive samples is better than that of RF, and the F1 value is the overall evaluation of the specificity and the sensitivity, so that the ACO-DBN model has the maximum F1 value and has the strongest correct prediction capability on the positive and negative real samples under the condition that the ACO-DBN model and the ACO-DBN model are excellent in performance. From FIG. 6, it can be seen that the curve of the ACO-DBN model is located above the RF and SVM values, and the corresponding AUC value is the maximum, which is 0.973. By combining the analysis, the ACO-DBN model shows better prediction performance and stability for landslide susceptibility evaluation, and the ACO-DBN model has great advantages and research values for processing complex nonlinear factors.
The following describes the landslide susceptibility evaluating apparatus provided by the present invention, and the landslide susceptibility evaluating apparatus described below and the landslide susceptibility evaluating method described above may be referred to with each other.
The landslide susceptibility evaluating apparatus of the present invention will be described below with reference to fig. 8, the apparatus comprising:
and the disaster-causing factor screening module 100 is used for screening out landslide susceptibility disaster-causing factors in the sample area. Due to the fact that the selection of the landslide susceptibility disaster-causing factors lacks theoretical basis and the landslide susceptibility disaster-causing factors possibly have strong correlation under different geographic environments, meanwhile, in a deep learning model, due to the fact that too large feature dimensionality easily causes reduction of model overfitting and execution efficiency, abnormal noise features of too many samples are learned, and the generalization capability of the model is also reduced, and therefore the landslide susceptibility disaster-causing factors need to be correspondingly screened. After the processing of step S100, the obtained landslide susceptibility disaster-causing factors are all weak-related geographical-environmental factors, i.e., weak-related features, which can exclude strong-related features, and at the same time, exclude features that contribute little to model training, so that the efficiency and accuracy of the subsequently generated model are improved.
The sample set establishing module 200 is configured to combine the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set.
The sample set establishing module 200 specifically includes collecting a landslide sample set of a sample area by using regional landslide survey, visual interpretation based on satellite images and intelligent identification based on remote sensing images, and the three devices complement and verify with each other. In this embodiment, the sample set creating module 200 also converts part of the surface landslide data into uniform point data (elements) by using Geographic Information System (Geo-Information System, GIS) software or the like.
The proneness evaluation module 300 is configured to use the disaster-causing factor-sample set as input data for training, and obtain an ant colony optimization algorithm-deep belief network model for generating a landslide proneness result of the to-be-evaluated area in a deep learning manner.
In the susceptibility evaluation module 300, an Ant Colony Optimization (ACO) -Deep Belief Network (DBN) model is adopted, and the DBN model is firstly applied to the landslide susceptibility evaluation by combining with the ACO in the landslide susceptibility evaluation device.
According to the landslide susceptibility evaluation device provided by the invention, the DBN model and the ACO are combined and applied to landslide susceptibility evaluation through the susceptibility evaluation module 300, the problems of model overfitting, slow convergence and the like are solved through the combination and parameters in the optimization strategy of the ACO optimization DBN model, the landslide susceptibility disaster-causing factor with weak correlation characteristics is preferably selected through the disaster-causing factor screening module 100, so that deeper relation between the landslide susceptibility disaster-causing factor and a sample is realized, and finally more values of the deep learning model in landslide susceptibility and other disaster prevention and reduction applications are realized.
In the following, the landslide susceptibility assessment apparatus of the present invention is described with reference to fig. 9, and the disaster-causing factor screening module 100 specifically includes:
and the acquisition unit 110 is used for acquiring landslide information of the sample area. The landslide information comprises remote sensing images, satellite images, basic geography, geology, terrain and geomorphology, meteorological hydrology, forest land types, soil types, historical landslide disaster records and landslide research literature data, and the remote sensing images, the satellite images, the basic geography, geology, the terrain and geomorphology, the meteorological hydrology, the forest land types and the soil types are basic data.
And the screening unit 120 is configured to screen out the landslide susceptibility disaster-causing factor corresponding to the landslide information from the landslide information by using the pearson correlation coefficient and the random forest feature importance ranking.
In the screening unit 120, for different geographic environments, the selection of the landslide susceptibility disaster-causing factor lacks a theoretical basis, and the disaster-causing factor may have strong correlation, and meanwhile, in the deep learning model, too large feature dimension easily causes reduction in model overfitting and execution efficiency, and abnormal noise features of too many samples are learned, which also reduces generalization capability of the model, and therefore, the landslide susceptibility disaster-causing factor needs to be screened. And screening the disaster-causing factors through the Pearson correlation coefficient and the importance ranking of the random forest features so as to improve the operation efficiency of the model and ensure the accuracy. The pearson correlation coefficient characterizes the degree of correlation between two evaluation factors by a specific value between-1 and 1. When the two evaluation factors are independent from each other and have no correlation, the corresponding correlation coefficient value is 0; when the two evaluation factors accord with the linear positive correlation, the correlation coefficient value between the two evaluation factors is 1; when the correlation coefficient values of the two evaluation factors are-1, it is indicated that a linear negative correlation relationship is satisfied between the factors. The random forest feature importance ordering is to directly measure the influence of each feature on the model prediction accuracy according to random forests, and the basic idea is to rearrange the sequence of a certain list of feature values and observe the accuracy of the model reduced. By the two device modes, not only can the strongly relevant characteristics be eliminated, but also the obtained landslide susceptibility disaster-causing factors are the weakly relevant geographical-environmental factors, namely the weakly relevant characteristics, and the characteristics with small contribution to model training are eliminated, so that the efficiency and the precision of the ACO-DBN model are improved.
After the processing of the acquisition unit 110 and the screening unit 120, in the landslide susceptibility evaluation device of the present invention, 9 screened landslide susceptibility disaster-inducing factors are respectively lithology, slope, distance from road, Normalized Difference Vegetation Index (NDVI), slope position, distance from water system, land utilization, curvature, and slope direction, and it can be understood that the above 9 landslide susceptibility disaster-inducing factors participate in the construction process of the ACO-DBN model.
Fig. 10 illustrates a physical structure diagram of an electronic device, and as shown in fig. 10, the electronic device may include: a processor (processor)810, a communication Interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication Interface 820 and the memory 830 communicate with each other via the communication bus 840. The processor 810 may invoke logic instructions in the memory 830 to perform a landslide susceptibility assessment method comprising the steps of:
s100, screening out landslide susceptibility disaster-causing factors in a sample area;
s200, combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
s300, the disaster-causing factor-sample set is used as input data for training, and an ant colony optimization algorithm-deep belief network model for generating landslide proneness results of the area to be evaluated is obtained in a deep learning mode.
In addition, the logic instructions in the memory 830 may be implemented in software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, which when executed by a computer, enable the computer to perform the landslide susceptibility assessment method provided by the above methods, the method comprising the steps of:
s100, screening out landslide susceptibility disaster-causing factors in a sample area;
s200, combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
s300, the disaster-causing factor-sample set is used as input data for training, and an ant colony optimization algorithm-deep belief network model for generating landslide proneness results of the area to be evaluated is obtained in a deep learning mode.
In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method for performing landslide liability assessment provided above, the method comprising the steps of:
s100, screening out landslide susceptibility disaster-causing factors in a sample area;
s200, combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
s300, the disaster-causing factor-sample set is used as input data for training, and an ant colony optimization algorithm-deep belief network model for generating landslide proneness results of the area to be evaluated is obtained in a deep learning mode.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A landslide susceptibility assessment method is characterized by comprising the following steps:
screening out landslide susceptibility disaster-causing factors in the sample area;
combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
and taking the disaster-causing factor-sample set as input data for training, and obtaining an ant colony optimization algorithm-deep belief network model for generating a landslide proneness result of the area to be evaluated in a deep learning mode.
2. The landslide susceptibility assessment method according to claim 1, wherein screening out landslide susceptibility disaster-causing factors specifically comprises the steps of:
collecting landslide information of a sample area; the landslide information comprises remote sensing images, satellite images, basic geography, geology, landform, meteorological hydrology, forest land types, soil types, historical landslide disaster records and landslide research literature data;
and screening the landslide susceptibility disaster-causing factors corresponding to the landslide information from the landslide information by using the Pearson correlation coefficient and the importance ranking of random forest features.
3. The landslide susceptibility assessment method of claim 2 wherein the collection of landslide samples taken from said sample area specifically comprises the steps of:
a landslide sample set of sample areas is collected using regional landslide investigation, visual interpretation based on satellite images, and intelligent identification based on remote sensing imagery.
4. The landslide susceptibility assessment method of claim 3 wherein collecting sample area collected landslide sample is converted from areal data to point data using area landslide survey, visual interpretation based on satellite images and intelligent identification based on remote sensing imagery.
5. The landslide susceptibility assessment method according to claim 1, wherein a heuristic ant colony algorithm is used to find the best path and iterate continuously during the training process of the ant colony optimization algorithm-deep belief network model to find the best combination and the best parameters of the ant colony optimization algorithm-deep belief network model until the iteration times are maximum, so as to obtain the ant colony optimization algorithm-deep belief network model with the best combination and parameters.
6. The landslide susceptibility assessment method according to claim 5, wherein in the parameters of the ant colony optimization algorithm-deep belief network model, the activation function is a hyperbolic tangent function, the optimizer is an adaptive moment estimation optimizer, the optimization strategy is a batch standardization optimization strategy, a ridge regression weight attenuation term regularization and early stop method is used, and the loss function is a cross entropy and a ridge regression adjustment term.
7. The landslide susceptibility assessment method of claim 5, wherein the architecture of the ant colony optimization algorithm-deep belief network model comprises an input layer, an output layer and four layers of restricted Boltzmann machine layers located between the input layer and the output layer, wherein the number of neurons in the four layers of restricted Boltzmann machine layers decreases step by step from the input layer to the output layer.
8. A landslide susceptibility assessment apparatus comprising:
the disaster-causing factor screening module (100) is used for screening out landslide susceptibility disaster-causing factors in the sample area;
the sample set establishing module (200) is used for combining the landslide susceptibility disaster-causing factor with a landslide sample set collected from the sample area to obtain a disaster-causing factor-sample set;
and the proneness evaluation module (300) is used for taking the disaster-causing factor-sample set as input data used for training and obtaining an ant colony optimization algorithm-deep belief network model used for generating a landslide proneness result of the area to be evaluated in a deep learning mode.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the landslide susceptibility assessment method according to any one of claims 1-7 are implemented when the program is executed by the processor.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the steps of the landslide susceptibility assessment method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110698085.5A CN113822522A (en) | 2021-06-23 | 2021-06-23 | Landslide susceptibility assessment method, device and equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110698085.5A CN113822522A (en) | 2021-06-23 | 2021-06-23 | Landslide susceptibility assessment method, device and equipment and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113822522A true CN113822522A (en) | 2021-12-21 |
Family
ID=78924001
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110698085.5A Pending CN113822522A (en) | 2021-06-23 | 2021-06-23 | Landslide susceptibility assessment method, device and equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113822522A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115830804A (en) * | 2022-10-24 | 2023-03-21 | 北京中地华安科技股份有限公司 | Intelligent early warning negative sample sampling method for pipeline geological disasters under constraint of easily-issued subareas |
| CN116526568A (en) * | 2023-07-03 | 2023-08-01 | 国网北京市电力公司 | Method, system, equipment and medium for optimizing distributed power supply of alternating current/direct current power distribution network |
| CN116881652A (en) * | 2023-06-26 | 2023-10-13 | 成都理工大学 | Landslide vulnerability evaluation method based on optimal negative sample and random forest model |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109063247A (en) * | 2018-06-26 | 2018-12-21 | 西安工程大学 | Landslide disaster forecasting procedure based on deepness belief network |
| JP2019085712A (en) * | 2017-11-02 | 2019-06-06 | 国際航業株式会社 | Slope stability determination assistance system |
| CN110346518A (en) * | 2019-07-25 | 2019-10-18 | 中南大学 | A kind of traffic emission pollution visualization method for early warning and its system |
| CN110705774A (en) * | 2019-09-26 | 2020-01-17 | 汉纳森（厦门）数据股份有限公司 | Vehicle energy consumption analysis prediction method and system |
| KR20200052398A (en) * | 2018-10-23 | 2020-05-15 | 강원대학교산학협력단 | Method and apparatus for landslide susceptibility mapping using machine-learning architecture |
| CN111898315A (en) * | 2020-07-27 | 2020-11-06 | 中国科学院空天信息创新研究院 | Landslide susceptibility assessment method based on fractal-machine learning hybrid model |
| CN112949189A (en) * | 2021-03-07 | 2021-06-11 | 湖南安全技术职业学院（长沙煤矿安全技术培训中心） | Modeling method for multi-factor induced landslide prediction based on deep learning |
-
2021
- 2021-06-23 CN CN202110698085.5A patent/CN113822522A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2019085712A (en) * | 2017-11-02 | 2019-06-06 | 国際航業株式会社 | Slope stability determination assistance system |
| CN109063247A (en) * | 2018-06-26 | 2018-12-21 | 西安工程大学 | Landslide disaster forecasting procedure based on deepness belief network |
| KR20200052398A (en) * | 2018-10-23 | 2020-05-15 | 강원대학교산학협력단 | Method and apparatus for landslide susceptibility mapping using machine-learning architecture |
| CN110346518A (en) * | 2019-07-25 | 2019-10-18 | 中南大学 | A kind of traffic emission pollution visualization method for early warning and its system |
| CN110705774A (en) * | 2019-09-26 | 2020-01-17 | 汉纳森（厦门）数据股份有限公司 | Vehicle energy consumption analysis prediction method and system |
| CN111898315A (en) * | 2020-07-27 | 2020-11-06 | 中国科学院空天信息创新研究院 | Landslide susceptibility assessment method based on fractal-machine learning hybrid model |
| CN112949189A (en) * | 2021-03-07 | 2021-06-11 | 湖南安全技术职业学院（长沙煤矿安全技术培训中心） | Modeling method for multi-factor induced landslide prediction based on deep learning |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115830804A (en) * | 2022-10-24 | 2023-03-21 | 北京中地华安科技股份有限公司 | Intelligent early warning negative sample sampling method for pipeline geological disasters under constraint of easily-issued subareas |
| CN115830804B (en) * | 2022-10-24 | 2023-08-22 | 北京中地华安科技股份有限公司 | Pipeline geological disaster intelligent early warning negative sample sampling method under constraint of easily-generated partition |
| CN116881652A (en) * | 2023-06-26 | 2023-10-13 | 成都理工大学 | Landslide vulnerability evaluation method based on optimal negative sample and random forest model |
| CN116881652B (en) * | 2023-06-26 | 2024-04-05 | 成都理工大学 | Landslide vulnerability evaluation method based on optimal negative sample and random forest model |
| CN116526568A (en) * | 2023-07-03 | 2023-08-01 | 国网北京市电力公司 | Method, system, equipment and medium for optimizing distributed power supply of alternating current/direct current power distribution network |
| CN116526568B (en) * | 2023-07-03 | 2023-09-15 | 国网北京市电力公司 | Method, system, equipment and medium for optimizing distributed power supply of alternating current/direct current power distribution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10360517B2 (en) | Distributed hyperparameter tuning system for machine learning | |
| KR102077804B1 (en) | Method and system for pre-processing machine learning data | |
| CN113822522A (en) | Landslide susceptibility assessment method, device and equipment and readable storage medium | |
| US7792769B2 (en) | Apparatus and method for learning and reasoning for systems with temporal and non-temporal variables | |
| US20200311556A1 (en) | Process and System Including an Optimization Engine With Evolutionary Surrogate-Assisted Prescriptions | |
| US10963802B1 (en) | Distributed decision variable tuning system for machine learning | |
| CN110751289B (en) | Online learning behavior analysis method based on Bagging-BP algorithm | |
| CN111582538A (en) | Community value prediction method and system based on graph neural network | |
| Bi et al. | A genetic algorithm-assisted deep learning approach for crop yield prediction | |
| Khosravi et al. | Performance Evaluation of Machine Learning Regressors for Estimating Real Estate House Prices | |
| Sánchez et al. | Obtaining transparent models of chaotic systems with multi-objective simulated annealing algorithms | |
| Bergen et al. | A review of supervised learning methods for classifying animal behavioural states from environmental features | |
| Raman et al. | Enhanced reptile search optimization with convolutional autoencoder for soil nutrient classification model | |
| CN116911994A (en) | External trade risk early warning system | |
| Adorada et al. | The comparison of feature selection methods in software defect prediction | |
| CN115936773A (en) | Internet financial black product identification method and system | |
| Ushio et al. | The application of deep learning to predict corporate growth | |
| Fitrianto et al. | Development of direct marketing strategy for banking industry: The use of a Chi-squared Automatic Interaction Detector (CHAID) in deposit subscription classification | |
| Thanekar et al. | Land Use Land Cover Prediction Analysis | |
| CN111340356A (en) | Method and apparatus for evaluating model interpretation tools | |
| Pohjalainen | Predicting service contract churn with decision tree models | |
| Widodo et al. | Finding the Best Performance of Bayesian and Naïve Bayes Models in Fraudulent Firms Classification through Varying Threshold | |
| Arumugam et al. | Performance evaluation of machine learning and deep learning techniques: a comparative analysis for house price prediction | |
| PP et al. | Predicting Student Performance Using Feature Selection Algorithms for Deep Learning Models | |
| Rostami et al. | Time Series Forecasting of House Prices: An evaluation of a Support Vector Machine and a Recurrent Neural Network with LSTM cells |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |