CN114792371A - Algorithm evaluation method, device, equipment and medium based on fuzzy hierarchical analysis - Google Patents
Algorithm evaluation method, device, equipment and medium based on fuzzy hierarchical analysis Download PDFInfo
- Publication number
- CN114792371A CN114792371A CN202210508826.3A CN202210508826A CN114792371A CN 114792371 A CN114792371 A CN 114792371A CN 202210508826 A CN202210508826 A CN 202210508826A CN 114792371 A CN114792371 A CN 114792371A
- Authority
- CN
- China
- Prior art keywords
- weight
- fuzzy
- algorithm
- feature
- features
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
- G06F18/2193—Validation; Performance evaluation; Active pattern learning techniques based on specific statistical tests
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The disclosure provides an algorithm evaluation method, device, equipment and medium based on fuzzy hierarchical analysis. The method comprises the following steps: establishing a hierarchical structure model based on a decision target, a candidate algorithm and characteristics; constructing a comparison matrix based on the hierarchical structure model and the triangular fuzzy scale, and determining fuzzy numbers and fuzzy geometric values corresponding to each feature; determining a fuzzy weight of the feature based on the fuzzy geometric value, and defuzzifying the fuzzy weight to obtain a weight value; normalizing the weight value to obtain the normalized weight of the characteristic; constructing a decision matrix of the candidate algorithm relative to each feature, and determining normalization weights corresponding to each candidate algorithm respectively by taking each feature as a standard; and calculating a weight result of each algorithm to be selected based on the normalization weight of the features and the normalization weight of the algorithm to be selected, and evaluating the algorithm to be selected based on the weight result. The method and the device can accurately and efficiently evaluate the candidate algorithm, reduce the evaluation cost and improve the reliability of the evaluation result.
Description
Technical Field
The present disclosure relates to the field of real-time segmentation and recognition algorithm technologies, and in particular, to an algorithm evaluation method, apparatus, device, and medium based on fuzzy hierarchical analysis.
Background
The real-time segmentation and recognition algorithm is widely used in a computer-aided system, is greatly helpful for detecting the fault of the lead, and can also obtain a very good effect on processing the streaming media data, and is one of the very important algorithms in the field of artificial intelligence at present. For example, in the field of smart power grids, real-time segmentation and identification of electric wires can be achieved by using a real-time segmentation and identification algorithm, however, the existing real-time segmentation and identification algorithms have a large number of schemes, and therefore, how to evaluate the real-time segmentation and identification algorithm selected by a user to determine the most efficient real-time segmentation and identification algorithm is one of the important concerns in the field.
In the prior art, when a real-time segmentation and recognition algorithm is evaluated, the following method is generally adopted, namely, various real-time segmentation and recognition algorithms to be evaluated are evaluated based on a traditional expert system, and when an algorithm evaluation index system is established by using the expert system, the weights of various indexes are calculated according to a large amount of basic data in an expert system database and by combining a model based on a convolutional neural network, so that various real-time segmentation and recognition algorithms are evaluated according to the index weights. The existing mode for evaluating the real-time segmentation and recognition algorithm has the problems of low evaluation efficiency, high evaluation cost, low accuracy of evaluation results, incapability of determining the most effective solution based on the evaluation results and the like.
Disclosure of Invention
In view of this, embodiments of the present disclosure provide an algorithm evaluation method, apparatus, device, and medium based on fuzzy hierarchical analysis, so as to solve the problems of low evaluation efficiency, high evaluation cost, and low accuracy of an evaluation result in the prior art.
In a first aspect of the embodiments of the present disclosure, an algorithm evaluation method based on fuzzy hierarchical analysis is provided, including: acquiring characteristics for evaluating the algorithm to be selected, and establishing a hierarchical structure model for evaluating the algorithm to be selected based on a preset decision target, the algorithm to be selected and the characteristics; constructing a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, performing fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometric value corresponding to each feature by using the fuzzy number; determining a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and performing defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature; judging the weight sum of the features, and normalizing the weight values of the features according to the judgment result to obtain normalization weights corresponding to the features respectively; respectively constructing a decision matrix of the candidate algorithm relative to each feature so as to determine a normalization weight corresponding to each candidate algorithm respectively by taking each feature as a standard time based on the decision matrix; and calculating a weight result of each algorithm to be selected based on the normalization weight corresponding to the features and the normalization weight corresponding to the algorithm to be selected, and evaluating the algorithm to be selected based on the weight result.
In a second aspect of the embodiments of the present disclosure, an algorithm evaluation device based on fuzzy hierarchical analysis is provided, including: the acquisition module is configured to acquire characteristics used for evaluating the algorithm to be selected, and establish a hierarchical structure model used for evaluating the algorithm to be selected based on a preset decision target, the algorithm to be selected and the characteristics; the construction module is configured to construct a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, perform fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculate a fuzzy geometric value corresponding to each feature by using the fuzzy number; the processing module is configured to determine a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and perform defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature; the judging module is configured to judge the total weight of the features and normalize the weight values of the features according to the judging result to obtain normalized weights corresponding to the features respectively; the determining module is configured to respectively construct a decision matrix of the candidate algorithms relative to each feature, so that normalization weights corresponding to each candidate algorithm respectively are determined by taking each feature as a standard based on the decision matrix; and the evaluation module is configured to calculate a weight result of each algorithm to be selected based on the normalization weight corresponding to the features and the normalization weight corresponding to the algorithm to be selected, and evaluate the algorithm to be selected based on the weight result.
In a third aspect of the embodiments of the present disclosure, an electronic device is provided, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method when executing the program.
In a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, in which a computer program is stored, which when executed by a processor implements the steps of the above-mentioned method.
The embodiment of the present disclosure adopts at least one technical scheme that can achieve the following beneficial effects:
establishing a hierarchical structure model for evaluating the candidate algorithm based on a preset decision target, the candidate algorithm and the characteristics by acquiring the characteristics for evaluating the candidate algorithm; constructing a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, performing fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometric value corresponding to each feature by using the fuzzy number; determining a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and performing defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature; judging the weight sum of the features, and normalizing the weight values of the features according to the judgment result to obtain normalization weights corresponding to the features respectively; respectively constructing a decision matrix of the candidate algorithm relative to each feature so as to determine a normalization weight corresponding to each candidate algorithm respectively by taking each feature as a standard based on the decision matrix; and calculating a weight result of each algorithm to be selected based on the normalization weight corresponding to the features and the normalization weight corresponding to the algorithm to be selected, and evaluating the algorithm to be selected based on the weight result. According to the technical scheme, the algorithm to be selected can be evaluated accurately and efficiently, the evaluation cost is reduced, and the reliability of the evaluation result is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings used in the embodiments or the prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without inventive efforts.
FIG. 1 is a schematic flow chart diagram of an algorithm evaluation method based on fuzzy hierarchical analysis according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a hierarchy model for algorithm evaluation provided by an embodiment of the disclosure;
fig. 3 is a schematic structural diagram of an algorithm evaluation device based on fuzzy hierarchical analysis according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of an electronic device provided in an embodiment of the present disclosure.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular system structures, techniques, etc. in order to provide a thorough understanding of the disclosed embodiments. It will be apparent, however, to one skilled in the art that the present disclosure may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present disclosure with unnecessary detail.
In a computer-aided system, real-time device segmentation is an indispensable module in the computer-aided system, and real-time segmentation and recognition algorithms used in real-time device segmentation are also widely applied in various fields. For example, in the streaming media data segmentation technology, a real-time segmentation and identification algorithm is used for segmenting and identifying streaming media data on a website, and for example, a real-time segmentation and identification algorithm of an electric wire in a mobile terminal is widely used in the field of smart power grids. Several existing real-time segmentation and recognition algorithms are described below with reference to specific scenes and contents, which may specifically include the following contents:
1) among the real-time non-contact Fall detection systems RT-Fall based on commercial WiFi devices, RT-Fall is an immediate, non-contact, cheaper but correct indoor Fall identification system using commercial WiFi devices. The RT-Fall utilizes the phase and amplitude of accessible fine-grained Channel State Information (CSI) in commercial WiFi equipment, achieves the purposes of real-time automatic segmentation and Fall detection for the first time, allows a user to perform daily activities, and does not need to wear any equipment on the body. Firstly, in this fall detection system, the CSI phase difference on the two antennas is a more sensitive basis signal than amplitude for activity identification, which can very reliably partition falls and fall-like activities. Secondly, a rapid power distribution descending mode of falling in a time-frequency domain is found, and new feature extraction and accurate falling segmentation and detection are further carried out. Actual results in four different indoor situations show that the average sensitivity of RT-Fall is improved by 14% and the specificity is improved by 10%. (references: H.Wang, D.Zhang, Y.Wang, J.Ma, Y.Wang, and S.Li, "RT-Fall: A real-time and contact failure detection system with communication WiFi devices," IEEE Transactions on Mobile Computing, vol.16, No.2, pp.511-526,2016.)
2) A real-time three-dimensional obstacle detection method includes the steps of covering coordinates of point cloud data by using an external parameter matrix, supplementing missing data in original data through an interpolation algorithm, eliminating the ground by using a RANSAC algorithm, and performing cluster segmentation on distance thresholds of a point cloud fusion segmenter and angle thresholds of different distance intervals by using an enhanced DBSCAN clustering algorithm. The detection method eliminates the missing rate and improves the actual performance. Finally, practical results show that this method generates good quality object detection online through NVIDIA Jetson TX 2. (references: Y.Liu, H.Zhu, B.Zhung, and Y.Su, "Real-Time Segmentation of spark 3D Point cloud on a TX2," DESSTECH Transactions on Computer Science Engineering, No. cisnr,2020.)
3) The advanced deep learning method is used for accurate and real-time segmentation of high-resolution intravascular ultrasound images, and the research aims to innovate and verify a deep learning program which can automatically correct segmentation of intravascular ultrasound (IVUS) image arrangement in real time. The developed deep learning method was validated on 20 vessels using the estimates of two expert analysts as reference criteria, the innovative deep learning method performed correctly, high resolution IVUS datasets could be segmented, and these methods could improve the use of IVUS in clinical trials and analysis. (reference: R.Bajaj et al., "Advanced deep learning method for acuate, real-time segmentation of high-resolution interactive ultrasound images," International Journal of diagnosis, vol.339, pp.185-191,2021.)
4) This study proposed an effective image-based fault detection system that can automatically find faults on the surface of a wire by using image processing methods and pattern recognition, and damage detection algorithms combining image improvement methods with principal component analysis algorithms. The analysis uses image improvement methods and denoising strategies to improve the overall value of the image. The proposed damage detection algorithm was validated by laboratory testing of 3 wires. The test results show that the proposed system can be used to diagnose a broken link. (reference: H. -N.Ho., K. -D.Kim, Y. -S.park, and J. -J.Lee. "An effective image-based data detection for a table surface in a table structure," Ndt E International, vol.58, pp.18-23,2013.)
In the field of smart power grids, real-time segmentation and identification algorithms are also used for automatically detecting and removing faults of electric wires, for example, the real-time segmentation and identification algorithms are applied to electric wires of mobile terminals, and the real-time segmentation and identification algorithms can be used for quickly and effectively finding out electric wire faults and identifying reasons and positions of the electric wire faults. Existing real-time wire segmentation and identification algorithms combine machine learning and artificial intelligence-based frameworks, so wire fault detection has become very advanced and efficient. Several known real-time wire segmentation and identification algorithms are described below with reference to the following references, which may include the following:
1) this article proposes a segmentation method based on live lines, which makes use of a comparison of equivalent object contours. This technique speeds up the segmentation mechanism, automatically converting anchor points that segment the reference contours into undivided target slices. Segmentation is enhanced using automatic contour correction so that 51% to 73% of the interaction time can be maintained, making the live-line value of segmentation the same as before. (reference: M.
J.Ehrhardt,and H.Handels,"Live-wire-based segmentation using similarities between corresponding image structures,"Computerized Medical Imaging Graphics,vol.31,no.7,pp.549-560,2007.)
2) Malmberg et al utilize a system that combines stereographic and tactile, simplifying the effectiveness of 3D collaboration. For the segmentation of the volume image, the research proposes a new method based on a two-dimensional live line method. The proposed method comprises two parts: an interface for rendering 3D live line arcs at the edges of objects in a volumetric image, and an algorithm that links 2 such arcs to form different surfaces. (reference: F.Malmberg, E.Vidholm, and I.
"A 3D live-wire segmentation method for volume images using haptic interaction,"in International Conference on Discrete Geometry for Computer Imagery,2006,pp.663-673:Springer.)
3) This study proposed a new framework called Active Geometry Function (AGF) to solve the complex problem of real-time segmentation. AGF has the advantage of mathematical efficiency and computational complexity, as well as many reliable features similar to level set profiles. The AGF can be divided in real time within a few milliseconds per frame, the dividing result is equal to the result of manual searching, and the performance is equal to that of manual searching. This ability to actually segment in real-time simplifies diagnosis and workflow, and also allows innovative requests to be made, such as collaborative image acquisition of interventional assistance and virtual segmentation. In applications involving online handwriting recognition, the actual performance is critical, however, conventional techniques typically wait until the entire arc is found before beginning the survey, which certainly creates a break in the recognition mechanism. (references: Q.Duan, E.D.Angelini, and A.F.Lane, "Real-time segmentation by active geometry functions," Computer methods programs in biomedicine, vol.98, No.3, pp.223-230,2010.)
4) Kour and Saabne propose an online Arabic character segmentation method based on real-time recognition. It defines some activities that waste time in writing, which is crucial for the segmentation mechanism. The proposed system has been structured and tested using the ADAB database with good results. (reference materials: G.Kour and R.Saaben, "Real-time segmentation of on-line hand writing absorbent description," in 201414 th International Conference on Frontiers in hand writing absorbent, 2014, pp.417-422: IEEE.)
Based on the above disclosure, the existing real-time wire segmentation and identification algorithms are relatively numerous, and therefore, the existing real-time wire segmentation and identification algorithms need to be evaluated to determine the most efficient and best-performing algorithm. The embodiment of the disclosure provides an algorithm evaluation method based on Fuzzy Analytic Hierarchy Process (FAHP), which evaluates a real-time segmentation and recognition algorithm by adopting a FAHP. FAHP is a simple and intuitive method for computing the weights of alternatives, criteria, and ranking them to determine the most efficient solution. The disclosed embodiments rank multiple alternatives (i.e., candidate algorithms) using a FAHP approach with fuzzy geometric means, which is typically used for multi-decision scenarios involving ambiguity and uncertainty.
The traditional method for solving the problem of multi-index decision evaluation is an Analytic Hierarchy Process (AHP), and generally comprises the steps of establishing a hierarchical structure model, constructing a judgment matrix, performing hierarchical single ordering and consistency check, performing hierarchical total ordering and consistency check and the like, and finally performing weighted summation based on the relative weight of each index and the normalized value of each index to finally obtain the comprehensive evaluation value of the system. The application range is wide, but the following defects still exist: the fuzziness of expert experience knowledge is not considered, and the scientificity of index weight cannot be guaranteed; the implicit internal relation of each index is not considered, the traditional AHP method only compares the importance of the indexes at the same level, but cannot consider the influence relation between the cross-category bottom indexes among the cross-level indexes; information can be integrated only after the evaluation of each evaluator is finished, and information integration cannot be performed when a complementary judgment matrix is constructed.
The biggest problem of the analytic hierarchy process AHP is that when a certain evaluation index of a hierarchy is many (such as more than four), the thinking consistency is difficult to guarantee. In this case, a Fuzzy Analytic Hierarchy Process (FAHP) which combines the advantages of the fuzzy analytic and the analytic hierarchy processes can solve the problem well. The basic idea and steps of the fuzzy analytic hierarchy process FAHP are basically the same as the steps of the AHP, but the two aspects are different:
1. the established judgment matrixes are different: in AHP, a judgment consistency matrix is established through pairwise comparison of elements, and in FAHP, a fuzzy consistency judgment matrix is established through pairwise comparison of elements;
2. the method of weighting the relative importance of each element in the matrix differs: the FAHP improves the problems of the traditional analytic hierarchy process and improves the decision reliability. FAHP is based on fuzzy numbers, and fuzzy consistency matrix.
In order to realize the evaluation of the real-time segmentation and recognition algorithm, the embodiment of the disclosure applies an FAHP method to measure the weights of standard factors and alternative schemes and sequences the weights so as to determine the most efficient and effective algorithm, and simultaneously considers the effectiveness and efficiency of the real-time segmentation and recognition algorithm.
Fig. 1 is a schematic flowchart of an algorithm evaluation method based on fuzzy hierarchical analysis according to an embodiment of the present disclosure. The algorithm evaluation method based on fuzzy hierarchy analysis of fig. 1 may be performed by a mobile terminal or a server. As shown in fig. 1, the algorithm evaluation method based on fuzzy hierarchical analysis may specifically include:
s101, obtaining characteristics for evaluating the candidate algorithm, and establishing a hierarchical structure model for evaluating the candidate algorithm based on a preset decision target, the candidate algorithm and the characteristics;
s102, constructing a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, carrying out fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometric value corresponding to each feature by using the fuzzy number;
s103, determining a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and performing defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature;
s104, judging the weight sum of the features, and normalizing the weight values of the features according to the judgment result to obtain normalized weights corresponding to the features respectively;
s105, respectively constructing a decision matrix of the candidate algorithm relative to each feature, so as to determine a normalization weight corresponding to each candidate algorithm respectively by taking each feature as a standard based on the decision matrix;
and S106, calculating a weight result of each algorithm to be selected based on the normalization weight corresponding to the features and the normalization weight corresponding to the algorithm to be selected, and evaluating the algorithm to be selected based on the weight results.
In particular, multiple evaluation indexes are involved in algorithm evaluation, so that the multi-index decision evaluation problem for algorithm evaluation is full of ambiguity and uncertainty, and the FAHP method is an advanced version of the AHP method based on the fuzzy logic theory. The FAHP method comprises a membership function and a conversion function from an AHP scale number to a fuzzy number, and is mostly used for multi-index decision evaluation problems. In addition, the FAHP technology mainly solves the problems of filling with ambiguity and uncertainty, resulting in difficult decision making, and in the disclosed embodiment, the FAHP method is used to evaluate real-time wire segmentation and identification algorithms in the mobile terminal.
Further, the decision target of the embodiment of the present disclosure refers to a problem to be solved by the FAHP method, that is, a target to be achieved by the FAHP method, and the decision target is an evaluation real-time segmentation and recognition algorithm in the embodiment of the present disclosure; the candidate algorithm is an alternative scheme in an FAHP method, namely various methods adopted for solving the problems, and is a real-time wire segmentation and identification algorithm in the embodiment of the disclosure; the feature refers to a standard factor in the FAHP method, that is, a factor, a standard, or a condition to be considered when evaluating a real-time segmentation and recognition algorithm, and may also be considered as a decision criterion, so in the following embodiments of the present disclosure, the feature may also be replaced by another term, for example, the standard, the factor, the condition, the criterion, and the like, and the replacement on the term does not form a limitation on the technical solution of the present disclosure.
Further, the features of the embodiments of the present disclosure are features obtained by means of feature extraction, raw data are obtained from relevant documents of existing real-time wire segmentation and identification algorithms, and the raw data of the reference documents are converted into digital features that can be processed while retaining the information of the raw data. The features are extracted from the raw data, some common features are selected from the extracted features, and the common features are used as the evaluation criteria of the algorithm. As shown in table 1, are features extracted from raw data for embodiments of the present disclosure.
TABLE 1 extracted features
In table 1, the numbers indicate the numbers corresponding to the references, and common features are selected as evaluation criteria from all the features extracted in table 1, for example, sensitivity, specificity, quality, time, consistency, and stability may be selected as criteria.
In some embodiments, establishing a hierarchical model for evaluating a candidate algorithm based on preset decision objectives, the candidate algorithm, and features includes: respectively taking the decision target as a highest layer, taking the algorithm to be selected as a lowest layer, taking the characteristics as an intermediate layer, and establishing a hierarchical structure model based on the highest layer, the lowest layer and the intermediate layer; the method comprises the steps of evaluating an algorithm to be selected as a decision target, wherein the algorithm to be selected comprises a plurality of algorithms to be evaluated, and the characteristics comprise a plurality of standard factors for evaluating the algorithm to be selected.
Specifically, after determining the features, decision targets and alternative schemes (i.e. candidate algorithms) in the algorithm evaluation, establishing a hierarchical structure model by using the features, decision targets and candidate algorithms based on an FAHP method; fig. 2 is a schematic diagram of a hierarchical structure model for algorithm evaluation provided by an embodiment of the present disclosure. As shown in fig. 2, the hierarchical structure model for algorithm evaluation may specifically include:
the highest layer in the hierarchical structure model is a decision target layer, and the decision target layer comprises a decision target for algorithm evaluation, namely, a real-time segmentation and recognition algorithm is evaluated; the bottom layer in the hierarchical structure model is a scheme layer, and the scheme layer comprises three alternative schemes, namely an algorithm-1, an algorithm-2 and an algorithm-3; an intermediate layer in the hierarchical structure model is a standard layer, which is also called a factor layer, a criterion layer, a feature layer and the like, the intermediate layer contains some criteria for algorithm evaluation, and the following six features are adopted as criteria: sensitivity, specificity, quality, time, consistency and stability.
In some embodiments, constructing a comparison matrix based on the hierarchical model and a predetermined triangular fuzzy scale comprises: and constructing a comparison matrix between the features by utilizing a triangular fuzzy scale based on all the features in the hierarchical structure model, wherein the triangular fuzzy scale is used for representing the importance degree between any two features.
Specifically, by decomposing, classifying problems and arranging them into a hierarchical framework, an Analytic Hierarchy Process (AHP) has been proposed to address the difficulties. The AHP method relies on expert opinion and thus the search decision is susceptible to subjectivity. The fuzzy hierarchy analysis process theory is expanded, and the AHP method and the FAHP method are used together. In the fuzzy analytic hierarchy process FAHP, the scale in the analytic hierarchy process AHP is converted into a fuzzy triangle scale, and table 3 shows the fuzzy scale.
TABLE 3 fuzzy scale
In the fuzzy scale of table 3, numbers 1-9 are used to indicate the importance of two factors to the decision making goal, and the fuzzy scale method is also called fuzzy scale or calibration method.
Further, a comparison matrix between the features is constructed by using a triangular fuzzy scale, and a paired comparison matrix n x n is drawn.
Based on the extracted features, a comparison matrix is constructed by using a triangular fuzzy scale, and the comparison matrix can also be called a judgment matrix or a pair comparison matrix. After the comparison matrix is constructed, different values are set for each factor pair in the comparison matrix, and the values are used for judging the importance of the decision target when two factors in the factor layer are compared.
TABLE 4 comparison matrix
In some embodiments, performing fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometric value corresponding to each feature by using the fuzzy number, includes: converting language numbers between the feature pairs in the comparison matrix into triangular fuzzy numbers by using a preset conversion formula to obtain triangular fuzzy numbers corresponding to each feature pair respectively, and converting the comparison matrix into a fuzzy pairing comparison decision matrix according to the triangular fuzzy numbers; and calculating the fuzzy geometric value corresponding to each feature by using a preset fuzzy geometric value calculation formula based on the triangular fuzzy number of each feature pair in the pairing comparison decision matrix.
Specifically, after setting a different value for each factor pair (feature pair) in the comparison matrix, each factor is given a fuzzy number, that is, a linguistic number between the feature pairs in the comparison matrix is converted into a triangular fuzzy number from either one of the following two formulas:
A -1 ＝(l,m,u) -1
A -1 ＝(1/u,1/m,1/l)
wherein l represents the lower digit, m represents the median, and u represents the upper digit.
Further, after the linguistic numbers are converted into fuzzy numbers (i.e., triangular fuzzy numbers) using the above formula, a fuzzy match-making comparison decision matrix is obtained, as shown in table 5.
TABLE 5 fuzzy pair comparison decision matrix
Further, based on the triangular fuzzy number of each feature pair in the pair comparison decision matrix, calculating a fuzzy geometric value corresponding to each feature through the following formula, wherein the fuzzy geometric value calculation formula is as follows:
where A represents the number of triangular ambiguities and n represents the number of features.
The blur geometry value for each criterion (i.e., each feature) can be calculated using the above formula, as shown in table 6.
TABLE 6 fuzzy geometry
Further, after the fuzzy geometric value of each standard is obtained, determining a fuzzy weight corresponding to each standard based on the fuzzy geometric value, and performing defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each standard; for the fuzzy weight W i The following formula is used for calculation:
W i ＝r i *(r 1 ,r 2 ,r 3 ,……,r 10 ) -1
wherein r is i The fuzzy geometry values for each standard are represented.
When the blurring process is performed on the blurring weight, the weight can be calculated by adopting the following formula:
(w i )＝l+m+u/3
wherein (w) i ) A weight value representing each criterion;
by performing the above-described averaging equation, the fuzzy weight W can be derived i Get the weight value (w) of each standard i )。
In some embodiments, judging the total weight of the features, and normalizing the weight values of the features according to the judgment result to obtain normalized weights corresponding to the features respectively, includes: calculating the weight sum of the weights corresponding to all the features, and when the weight sum is less than or equal to 1, directly taking the weight corresponding to each feature as the normalized weight of the feature; when the weight sum is larger than 1, converting the weight value of the feature into a normalization weight so as to enable the normalization weight sum corresponding to all the features to be 1; ranking the features based on the normalized weight corresponding to each feature.
Specifically, the weights of all the standards are summed, and if the sum of the weights is greater than 1, the weight values need to be converted into a normalized form, that is, the weight values are normalized, and the normalized weights can be calculated by using the following formula:
fuzzy geometry values, fuzzy weights, weights and normalized weights for each criterion are included as shown in table 7, which also shows the ranking for each criterion.
Further, a decision matrix of the candidate algorithm relative to each feature is respectively constructed, so that on the basis of the decision matrix, normalization weights corresponding to each candidate algorithm when each feature is used as a standard are determined. In practical applications, the above embodiment is to process an intermediate layer (i.e., a standard layer) of a hierarchical structure model, however, the FAHP method further needs to construct a judgment matrix of a scheme layer relative to the standard layer (i.e., a judgment matrix between a scheme and each standard), so as to obtain a corresponding scheme pairing decision matrix under each standard; the contents of the decision matrix for the solution pair under each standard are described in detail below with algorithm-1, algorithm-2, and algorithm-3 as alternatives. It should be noted that the calculation process of the scheme pairing decision matrix is similar to the calculation process of the fuzzy pairing comparison decision matrix and the fuzzy geometric value, the fuzzy weight, the weight and the normalization weight of the standard layer in the above embodiment, and is not described herein again.
Table 8 lists the weights and normalized weights for each selection scheme based on sensitivity.
TABLE 8 scheme pairing decision matrix based on sensitivity criteria
Table 9 lists the weights and normalized weights for each selection based on specificity.
TABLE 9 scheme pairing decision matrix based on specificity criteria
Table 10 lists the weights and normalized weights for each selection based on quality.
TABLE 10 quality criteria based scheme pairing decision matrix
Table 11 lists the weights and normalized weights for each selection scheme based on time.
TABLE 11 scheme pairing decision matrix based on time criteria
Table 12 lists the weights and normalized weights for each selection scheme based on consistency.
TABLE 12 scheme pairing decision matrix based on consistency criteria
Table 13 lists the weights and normalized weights for each selection based on stability.
TABLE 13 stability criteria based scheme pairing decision matrix
In some embodiments, calculating a weight result of each candidate algorithm based on the normalization weights corresponding to the features and the normalization weights corresponding to the candidate algorithms, and evaluating the candidate algorithms based on the weight results includes: and respectively multiplying the normalization weight of the features by the normalization weight of the algorithms to be selected, which is obtained by calculation with the features as standard time, to obtain a weight result corresponding to each algorithm to be selected, and ranking the algorithms to be selected according to the weight results so as to evaluate the algorithms to be selected according to the ranking.
Specifically, after the normalization weight corresponding to each standard and the normalization weight corresponding to the candidate algorithm under each standard are obtained by sequentially processing with an FAHP method, the weight result of each candidate algorithm is calculated by the following formula:
where m represents the number of alternatives. The normalized weights for each criterion (i.e., the criterion weights in table 14) are multiplied by each row by the above formula, and the results for each alternative are calculated and ranked.
TABLE 14 weight results and Algorithm ranking Using FAHP method
As can be seen from table 14, the decision result of the FAHP method for each candidate scheme is that the rank of the algorithm-2 is highest, the rank of the algorithm-1 is lowest, and the rank of the algorithm-3 is middle, so that it can be concluded that the algorithm-2 is the most efficient algorithm among the three candidate algorithms. In practical applications, the candidate algorithms of the embodiment of the present disclosure include an algorithm for real-time segmentation and identification of wires in the mobile terminal image.
Real-time segmentation is the activity of obtaining a computationally efficient real-time segmentation (while maintaining substantial accuracy). Real-time segmentation and recognition algorithms are widely used in this modern society, and real-time device segmentation is an essential module of computer-aided systems. These algorithms are very helpful and highly efficient for mobile terminal-based wire fault detection. In the disclosed embodiment, the real-time segmentation and recognition algorithm is evaluated using the FAHP method. The six criteria and three alternatives are weighted and ranked according to their value. Calculating the weight of each standard to obtain the highest weight of the sensitivity standard, wherein the weight is 0.220, and ranking is first; secondly, time, the weight value of 0.214 and the rank of the second is carried out; the weight value of the quality is 0.200, and the rank is third; the weight value of consistency is 0.132, and ranking is fourth; the weight value of the stability is 0.118, and the rank is fifth; and the weight of the specificity is the lowest, the weight value is 0.113, and the rank is the sixth. And then, sequentially weighing the weights of the alternative schemes (namely the candidate algorithms) according to each standard, wherein the result shows that the weight of the algorithm-2 is the highest and is 0.342, the weight of the algorithm-3 is 0.331, and the weight of the algorithm-1 is the lowest and is 0.322. In addition, the present disclosure effectively evaluates the real-time segmentation and recognition algorithms and discusses various uses of the real-time segmentation and recognition algorithms.
The following are embodiments of the disclosed apparatus that may be used to perform embodiments of the disclosed methods. For details not disclosed in the embodiments of the apparatus of the present disclosure, refer to the embodiments of the method of the present disclosure.
Fig. 3 is a schematic structural diagram of an algorithm evaluation device based on fuzzy hierarchical analysis according to an embodiment of the present disclosure. As shown in fig. 3, the algorithm evaluation device based on fuzzy hierarchy analysis includes:
the acquisition module 301 is configured to acquire features used for evaluating the algorithm to be selected, and establish a hierarchical structure model used for evaluating the algorithm to be selected based on a preset decision target, the algorithm to be selected and the features;
a constructing module 302 configured to construct a comparison matrix based on the hierarchical structure model and a predetermined triangular fuzzy scale, perform fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculate a fuzzy geometric value corresponding to each feature by using the fuzzy number;
the processing module 303 is configured to determine a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and perform defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature;
a judging module 304, configured to judge the total weight of the features, and normalize the weight values of the features according to the judgment result to obtain normalized weights corresponding to the features respectively;
a determining module 305 configured to construct a decision matrix of the candidate algorithm with respect to each feature, respectively, so as to determine a normalization weight corresponding to each candidate algorithm when each feature is used as a standard based on the decision matrix;
and the evaluation module 306 is configured to calculate a weight result of each candidate algorithm based on the normalization weight corresponding to the feature and the normalization weight corresponding to the candidate algorithm, and evaluate the candidate algorithm based on the weight result.
In some embodiments, the obtaining module 301 in fig. 3 sets the decision target as the highest layer, the candidate algorithm as the lowest layer, the feature as the middle layer, and a hierarchical structure model based on the highest layer, the lowest layer, and the middle layer; the method comprises the steps of evaluating an algorithm to be selected as a decision target, wherein the algorithm to be selected comprises a plurality of algorithms to be evaluated, and the characteristics comprise a plurality of standard factors for evaluating the algorithm to be selected.
In some embodiments, construction module 302 of FIG. 3 constructs a comparison matrix between features using a triangular blur scale based on all features in the hierarchy model, wherein the triangular blur scale is used to characterize the degree of importance between any two features.
In some embodiments, the constructing module 302 in fig. 3 converts the language numbers between the feature pairs in the comparison matrix into triangular fuzzy numbers by using a preset conversion formula, so as to obtain triangular fuzzy numbers corresponding to each feature pair, and converts the comparison matrix into a fuzzy match comparison decision matrix according to the triangular fuzzy numbers; and calculating the fuzzy geometric value corresponding to each feature by using a preset fuzzy geometric value calculation formula based on the triangular fuzzy number of each feature pair in the pairing comparison decision matrix.
In some embodiments, the determining module 304 in fig. 3 calculates a sum of weights corresponding to all features, and when the sum of weights is less than or equal to 1, directly takes the weight corresponding to each feature as a normalized weight of the feature; when the weight sum is larger than 1, converting the weight value of the feature into a normalized weight so as to enable the normalized weight sum corresponding to all the features to be 1; ranking the features based on the normalized weight corresponding to each feature.
In some embodiments, the evaluation module 306 in fig. 3 multiplies the normalized weights of the features by the normalized weights of the candidate algorithms calculated by using the features as the standard, so as to obtain a weight result corresponding to each candidate algorithm, and ranks the candidate algorithms according to the weight results, so as to evaluate the candidate algorithms according to the ranking.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by functions and internal logic of the process, and should not constitute any limitation to the implementation process of the embodiments of the present disclosure.
Fig. 4 is a schematic diagram of an electronic device 4 provided by the embodiment of the present disclosure. As shown in fig. 4, the electronic apparatus 4 of this embodiment includes: a processor 401, a memory 402 and a computer program 403 stored in the memory 402 and executable on the processor 401. The steps in the various method embodiments described above are implemented when the processor 401 executes the computer program 403. Alternatively, the processor 401 implements the functions of the respective modules/units in the above-described respective apparatus embodiments when executing the computer program 403.
The electronic device 4 may be a desktop computer, a notebook, a palm computer, a cloud server, or other electronic devices. The electronic device 4 may include, but is not limited to, a processor 401 and a memory 402. Those skilled in the art will appreciate that fig. 4 is merely an example of electronic device 4 and does not constitute a limitation of electronic device 4 and may include more or fewer components than shown, or different components.
The Processor 401 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, or the like.
The storage 402 may be an internal storage unit of the electronic device 4, for example, a hard disk or a memory of the electronic device 4. The memory 402 may also be an external storage device of the electronic device 4, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like provided on the electronic device 4. The memory 402 may also include both internal storage units of the electronic device 4 and external storage devices. The memory 402 is used for storing computer programs and other programs and data required by the electronic device.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules, so as to perform all or part of the functions described above. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit.
The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the present disclosure may implement all or part of the flow of the method in the above embodiments, and may also be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of the above methods and embodiments. The computer program may comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain suitable additions or additions that may be required in accordance with legislative and patent practices within the jurisdiction, for example, in some jurisdictions, computer readable media may not include electrical carrier signals or telecommunications signals in accordance with legislative and patent practices.
The above examples are only intended to illustrate the technical solutions of the present disclosure, not to limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, it should 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; such modifications and substitutions do not depart from the spirit and scope of the embodiments of the present disclosure, and they should be construed as being included in the scope of the present disclosure.
Claims (10)
1. An algorithm evaluation method based on fuzzy hierarchical analysis is characterized by comprising the following steps:
acquiring characteristics for evaluating a candidate algorithm, and establishing a hierarchical structure model for evaluating the candidate algorithm based on a preset decision target, the candidate algorithm and the characteristics;
constructing a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, performing fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometric value corresponding to each feature by using the fuzzy number;
determining a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and performing defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature;
judging the weight sum of the features, and normalizing the weight values of the features according to the judgment result to obtain normalized weights corresponding to the features respectively;
respectively constructing a decision matrix of the candidate algorithm relative to each feature, so as to determine a normalization weight corresponding to each candidate algorithm respectively by taking each feature as a standard time based on the decision matrix;
and calculating a weight result of each algorithm to be selected based on the normalization weight corresponding to the characteristic and the normalization weight corresponding to the algorithm to be selected, and evaluating the algorithm to be selected based on the weight result.
2. The method of claim 1, wherein the building a hierarchical model for evaluating the candidate algorithm based on preset decision objectives, the candidate algorithm, and the features comprises:
respectively taking the decision target as a highest layer, taking the algorithm to be selected as a lowest layer, taking the feature as a middle layer, and establishing the hierarchical structure model based on the highest layer, the lowest layer and the middle layer;
and evaluating the candidate algorithm as the decision target, wherein the candidate algorithm comprises a plurality of algorithms to be evaluated, and the characteristics comprise a plurality of standard factors for evaluating the candidate algorithm.
3. The method of claim 1, wherein constructing a comparison matrix based on the hierarchical model and a predetermined triangular fuzzy scale comprises:
and constructing a comparison matrix between the features by utilizing the triangular fuzzy scale based on all the features in the hierarchical structure model, wherein the triangular fuzzy scale is used for representing the importance degree between any two features.
4. The method according to claim 1, wherein the performing fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculating a fuzzy geometry value corresponding to each feature by using the fuzzy number comprises:
converting language numbers between the feature pairs in the comparison matrix into triangular fuzzy numbers by using a preset conversion formula to obtain triangular fuzzy numbers corresponding to each feature pair respectively, and converting the comparison matrix into a fuzzy pairing comparison decision matrix according to the triangular fuzzy numbers;
and calculating a fuzzy geometric value corresponding to each feature by using a preset fuzzy geometric value calculation formula based on the triangular fuzzy number of each feature pair in the pairing comparison decision matrix.
5. The method according to claim 1, wherein the determining the sum of the weights of the features and normalizing the weights of the features according to the determination result to obtain normalized weights corresponding to the features respectively comprises:
calculating the weight sum of the weights corresponding to all the features, and when the weight sum is less than or equal to 1, directly taking the weight corresponding to each feature as the normalized weight of the feature; when the weight sum is larger than 1, converting the weight value of the feature into a normalization weight so as to enable the normalization weight sum corresponding to all the features to be 1; ranking the features based on the normalized weight corresponding to each of the features.
6. The method according to claim 1, wherein the calculating a weight result of each candidate algorithm based on the normalization weight corresponding to the feature and the normalization weight corresponding to the candidate algorithm, and evaluating the candidate algorithm based on the weight results comprises:
and respectively multiplying the normalization weight of the characteristic by the normalization weight of the candidate algorithms calculated by taking the characteristic as a standard to obtain a weight result corresponding to each candidate algorithm, and ranking the candidate algorithms according to the weight results so as to evaluate the candidate algorithms according to the ranking.
7. The method according to any of claims 1-6, wherein the candidate algorithms include an algorithm for real-time segmentation and identification of wires in mobile terminal images.
8. An algorithm evaluation device based on fuzzy hierarchical analysis is characterized by comprising:
the system comprises an acquisition module, a selection module and a selection module, wherein the acquisition module is configured to acquire characteristics used for evaluating a to-be-selected algorithm and establish a hierarchical structure model used for evaluating the to-be-selected algorithm based on a preset decision target, the to-be-selected algorithm and the characteristics;
the construction module is configured to construct a comparison matrix based on the hierarchical structure model and a preset triangular fuzzy scale, perform fuzzy number calculation on the features in the comparison matrix to obtain a fuzzy number corresponding to each feature, and calculate a fuzzy geometric value corresponding to each feature by using the fuzzy number;
the processing module is configured to determine a fuzzy weight corresponding to each feature based on the fuzzy geometric value, and perform defuzzification processing on the fuzzy weight to obtain a weight value corresponding to each feature;
the judging module is configured to judge the total weight of the features and normalize the weight values of the features according to the judging result to obtain normalized weights corresponding to the features respectively;
the determining module is configured to respectively construct a decision matrix of the candidate algorithms relative to each feature, so that on the basis of the decision matrix, normalization weights corresponding to the candidate algorithms respectively are determined when each feature is taken as a standard;
and the evaluation module is configured to calculate a weight result of each candidate algorithm based on the normalization weight corresponding to the feature and the normalization weight corresponding to the candidate algorithm, and evaluate the candidate algorithm based on the weight result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method according to any one of claims 1 to 7 when executing the program.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210508826.3A CN114792371B (en) | 2022-05-10 | 2022-05-10 | Method, device, equipment and medium for determining wire real-time segmentation and identification algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210508826.3A CN114792371B (en) | 2022-05-10 | 2022-05-10 | Method, device, equipment and medium for determining wire real-time segmentation and identification algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114792371A true CN114792371A (en) | 2022-07-26 |
| CN114792371B CN114792371B (en) | 2023-07-25 |
Family
ID=82463671
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210508826.3A Active CN114792371B (en) | 2022-05-10 | 2022-05-10 | Method, device, equipment and medium for determining wire real-time segmentation and identification algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114792371B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009026589A2 (en) * | 2007-08-23 | 2009-02-26 | Fred Cohen | Method and/or system for providing and/or analizing and/or presenting decision strategies |
| CN104636447A (en) * | 2015-01-21 | 2015-05-20 | 上海天呈医流科技股份有限公司 | Intelligent evaluation method and system for medical instrument B2B website users |
| US20160239766A1 (en) * | 2015-02-16 | 2016-08-18 | Nathan R. Cameron | Systems, methods, and user interfaces for evaluating quality, health, safety, and environment data |
| CN107292534A (en) * | 2017-07-12 | 2017-10-24 | 国网福建省电力有限公司 | The yardstick competition evaluation method and device of urban power distribution network long term dynamics investment |
| CN109543737A (en) * | 2018-11-15 | 2019-03-29 | 国网四川省电力公司信息通信公司 | A kind of information system health degree appraisal procedure based on FAHP_FCA combination weighting |
| CN112164059A (en) * | 2020-10-20 | 2021-01-01 | 沈阳东软智能医疗科技研究院有限公司 | Focus detection method, device and related product |
| CN113592324A (en) * | 2021-08-05 | 2021-11-02 | 国网江苏省电力有限公司无锡供电分公司 | Cable terminal tower live-line work risk assessment method based on hierarchical analysis |
| CN113822538A (en) * | 2021-08-27 | 2021-12-21 | 长春工程学院 | AHP-CRITIC-based small and medium-sized earth and rockfill dam operation evaluation method |
| CN114266486A (en) * | 2021-12-24 | 2022-04-01 | 桂林电子科技大学 | AHP-FCE method-based risk evaluation method for whole-process cost consultation service |
-
2022
- 2022-05-10 CN CN202210508826.3A patent/CN114792371B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009026589A2 (en) * | 2007-08-23 | 2009-02-26 | Fred Cohen | Method and/or system for providing and/or analizing and/or presenting decision strategies |
| CN104636447A (en) * | 2015-01-21 | 2015-05-20 | 上海天呈医流科技股份有限公司 | Intelligent evaluation method and system for medical instrument B2B website users |
| US20160239766A1 (en) * | 2015-02-16 | 2016-08-18 | Nathan R. Cameron | Systems, methods, and user interfaces for evaluating quality, health, safety, and environment data |
| CN107292534A (en) * | 2017-07-12 | 2017-10-24 | 国网福建省电力有限公司 | The yardstick competition evaluation method and device of urban power distribution network long term dynamics investment |
| CN109543737A (en) * | 2018-11-15 | 2019-03-29 | 国网四川省电力公司信息通信公司 | A kind of information system health degree appraisal procedure based on FAHP_FCA combination weighting |
| CN112164059A (en) * | 2020-10-20 | 2021-01-01 | 沈阳东软智能医疗科技研究院有限公司 | Focus detection method, device and related product |
| CN113592324A (en) * | 2021-08-05 | 2021-11-02 | 国网江苏省电力有限公司无锡供电分公司 | Cable terminal tower live-line work risk assessment method based on hierarchical analysis |
| CN113822538A (en) * | 2021-08-27 | 2021-12-21 | 长春工程学院 | AHP-CRITIC-based small and medium-sized earth and rockfill dam operation evaluation method |
| CN114266486A (en) * | 2021-12-24 | 2022-04-01 | 桂林电子科技大学 | AHP-FCE method-based risk evaluation method for whole-process cost consultation service |
Non-Patent Citations (6)
| Title |
|---|
| PFATB D: "Fuzzy Analytical Hierarchy Process (FAHP) using geometric mean method to select best processing framework adequate to big data", 《JOURNAL OF THEORETICAL AND APPLIED INFORMATION TECHNOLOGY》, vol. 99, no. 1, pages 207 - 226 * |
| UDUPA J K等: "A framework for evaluating image segmentation algorithms", 《COMPUTERIZED MEDICAL IMAGING AND GRAPHICS》, vol. 30, no. 2, pages 75 - 87, XP024903627, DOI: 10.1016/j.compmedimag.2005.12.001 * |
| WANG L等: "Decision support system for real-time segmentation and identification algorithm for wires in mobile terminals using fuzzy AHP method", 《SOFT COMPUTING》, vol. 26, no. 20, pages 10915 - 10926 * |
| 侯格贤: "图象分割质量评价方法研究", 《中国图象图形学报》, vol. 5, no. 1, pages 39 - 43 * |
| 方志丹等: "基于层级识别模型的输电线路杆塔小金具缺陷识别方法", 《电力信息与通信技术》, vol. 18, no. 9, pages 16 - 24 * |
| 米红甫等: "基于FAHP的高层建筑火灾风险评价研究", 《武汉理工大学学报（信息与管理工程版）》, vol. 39, no. 2, pages 148 - 152 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114792371B (en) | 2023-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3961484A1 (en) | Medical image segmentation method and device, electronic device and storage medium | |
| CN110287932B (en) | Road blocking information extraction method based on deep learning image semantic segmentation | |
| CN109859245B (en) | Multi-target tracking method and device for video target and storage medium | |
| Putra et al. | Enhanced skin condition prediction through machine learning using dynamic training and testing augmentation | |
| CN110544233B (en) | Depth image quality evaluation method based on face recognition application | |
| CN106886980B (en) | Point cloud density enhancement method based on three-dimensional laser radar target identification | |
| CN112215101A (en) | Attention mechanism-based three-dimensional target identification method and system | |
| Liu et al. | SRAS‐net: Low‐resolution chromosome image classification based on deep learning | |
| Xiao et al. | Research on defect detection method of powder metallurgy gear based on machine vision | |
| CN115937552A (en) | Image matching method based on fusion of manual features and depth features | |
| CN117788810A (en) | Learning system for unsupervised semantic segmentation | |
| Jiang et al. | Pseudo‐Siamese residual atrous pyramid network for multi‐focus image fusion | |
| CN114792371B (en) | Method, device, equipment and medium for determining wire real-time segmentation and identification algorithm | |
| Du et al. | A scale‐sensitive heatmap representation for multi‐person pose estimation | |
| CN111624606A (en) | Radar image rainfall identification method | |
| Peng et al. | Improving the detection of the prostrate in ultrasound images using machine learning based image processing | |
| CN116310832A (en) | Remote sensing image processing method, device, equipment, medium and product | |
| Li et al. | Brown rice germ integrity identification based on deep learning network | |
| CN111833395B (en) | Direction-finding system single target positioning method and device based on neural network model | |
| CN115131384A (en) | Bionic robot 3D printing method, device and medium based on edge preservation | |
| CN114677530A (en) | Clustering algorithm effectiveness evaluation method, device and medium based on wavelet shape descriptor | |
| CN113436322A (en) | Three-dimensional reconstruction method, device, equipment and storage medium for fundus blood vessels | |
| Gonzalez-Huitron et al. | Jaccard distance as similarity measure for disparity map estimation | |
| Ji et al. | CellDet: Dual-Task Cell Detection Network for IHC-Stained Image Analysis | |
| Gorodissky et al. | Large field and high resolution: Detecting needle in haystack |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |