CN108959084B - Markov vulnerability prediction quantity method based on smoothing method and similarity - Google Patents
Markov vulnerability prediction quantity method based on smoothing method and similarity Download PDFInfo
- Publication number
- CN108959084B CN108959084B CN201810701155.6A CN201810701155A CN108959084B CN 108959084 B CN108959084 B CN 108959084B CN 201810701155 A CN201810701155 A CN 201810701155A CN 108959084 B CN108959084 B CN 108959084B
- Authority
- CN
- China
- Prior art keywords
- equal
- less
- vulnerability
- time node
- prediction
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3604—Software analysis for verifying properties of programs
- G06F11/3608—Software analysis for verifying properties of programs using formal methods, e.g. model checking, abstract interpretation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A Markov vulnerability prediction quantity method based on a smoothing method and similarity is characterized in that a security vulnerability is used as a research object, historical data information of the security vulnerability is inspected to form a vulnerability complete set, and the vulnerability complete set is reasonably divided into a direct prediction set and an indirect prediction set. And then, predicting the direct prediction set by using a Markov method improved by an exponential smoothing method, establishing the relation between the direct prediction set and the indirect prediction set by using a cosine similarity principle, and further predicting the indirect prediction set. And finally, integrating the prediction results of the direct prediction set and the indirect prediction set, and providing a high-accuracy prediction value for workers in related fields.
Description
Technical Field
The invention belongs to the technical field of computer information security, relates to an exponential smoothing method, cosine similarity and Markov algorithm, and particularly relates to a Markov vulnerability prediction quantity method based on the smoothing method and the similarity.
Background
The traditional software engineering subjects think that due to the principles of limited ability or insufficient experience of programmers, unreasonable software development process and the like, software inevitably has defects and hidden dangers. Among them, the defects and hidden dangers related to the security of the computer system are called security holes. With the rapid development of computer science, the attention of various industries to security vulnerabilities is increasing. Therefore, a method for predicting the number of the security vulnerabilities with high accuracy is provided for workers in the related field, and the method is significant work.
For quantitative prediction, the conventional method is based on statistical principles, and takes statistical indexes (such as arithmetic mean) as predicted values. Although this method is simple and easy, the correlation between different types of data is not considered, and therefore, it is difficult to obtain an accurate predicted value. The Markov algorithm is a modern prediction method, the algorithm fully considers the interconversion relationship among different types of data, and compared with the traditional method, the accuracy of prediction is greatly improved. However, how to determine the distribution of the state transition matrix more scientifically is a big problem in front of related researchers.
Disclosure of Invention
In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a method for predicting the number of holes in markov based on a smoothing method and a similarity, which improves the markov algorithm by using an exponential smoothing method and a cosine similarity. Firstly, a divide-and-conquer strategy is adopted, the average level of a whole set of the loopholes is taken as a reference, the loopholes which are closer to the loopholes are divided into a direct prediction set, and the loopholes which are farther from the loopholes are divided into an indirect prediction set. And secondly, investigating data distribution of various loopholes in the direct prediction set, and iteratively improving a state transition matrix and a probability matrix in a Markov method by using an exponential smoothing method to further obtain a predicted value of the direct prediction set. And finally, finding out the most similar vulnerability types for various vulnerabilities in the indirect prediction set in the direct prediction set by using cosine similarity, and carrying out proper scaling transformation by taking the predicted values as references to obtain the predicted values of the vulnerability number in the indirect prediction set.
In order to achieve the purpose, the invention adopts the technical scheme that:
a Markov prediction vulnerability quantity method based on a smoothing method and similarity is characterized by comprising the following steps:
(1) firstly, the historical quantity information of the security vulnerability is inspected to form a vulnerability complete set, and the vulnerability complete set is divided into a direct prediction set and an indirect prediction set
Examining historical quantity information of the security vulnerabilities to form a vulnerability complete set, recording the vulnerability complete set as U, wherein the U comprises all quantity information of n types of vulnerabilities in m time nodes, and recording the quantity informationuijRepresenting the number of the ith vulnerability in the jth time node, wherein i is more than or equal to 1 and less than or equal to n, j is more than or equal to 1 and less than or equal to m, and inspecting the average number level of the vulnerability complete set, namely calculating the arithmetic mean of all data in UAnd rounding the calculated arithmetic mean down:
note that the i-th vulnerability is UiAnd (3) inspecting the average quantity level of various loopholes, namely calculating the arithmetic mean of data of each row in the U, and rounding down:
wherein i is more than or equal to 1 and less than or equal to n,
to be provided withTaking 1 as an initial step length as a reference, p as a parameter for determining the range of the direct prediction set, generally, p is more than 0 and less than 1, and in order to ensure the accuracy of the prediction result, the value range of p is recommended to be more than or equal to 0.5 and less than 1,
the algorithm is as follows: algorithm for dividing vulnerability complete set into direct prediction set and indirect prediction set
Inputting: vulnerability complete set U and parameter p
(2) Predicting the number of various vulnerabilities in the direct prediction set:
the method for predicting the number of various vulnerabilities in the direct prediction set mainly comprises the following steps:
1) obtaining an actual state transition matrix
Setting w elements in a direct prediction set S, namely S contains all quantity information of w types of vulnerabilities on m time nodes; note the booksijThe number of the ith vulnerability in the jth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to m,
let QtRepresenting the actual state transition matrix from the (t-1) th time node to the t-th time nodeqijtRepresenting the actual probability of transferring the ith vulnerability into the jth vulnerability from the (t-1) th time node to the tth time node, wherein t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w;
wherein d isijThe number of times that the number of the ith loopholes is reduced and the number of the jth loopholes is increased is represented, i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to w, and j is not equal to i; f. ofijIndicates the ratio of the number of i-th bugs decreasing and the number of j-th bugs increasing to the number of i-th bugs decreasing, i.e. the number of1≤i≤w,1≤j≤w,j≠i;qitRepresents QtWherein i is not less than 1 and not more than w;
the algorithm is as follows: determining qitActual state transition matrix Q from t-1 time node to t time nodetRow i q ofitIs calculated by
Inputting: direct prediction set S, parameter fij,
And (3) outputting: q. q.sit
Wherein, the value of i is 1, 2, … … and w in sequence, and a complete actual state transition matrix Q can be obtainedtTaking the t values of 2, 3, … … and m in sequence to obtain all actual state transition matrixes;
2) obtaining a predicted state transition matrix
Is Q'tRepresenting the predicted state transition matrix from the (t-1) th time node to the t-th time nodeq’ijtRepresenting the prediction probability of transferring the ith vulnerability into the jth vulnerability from the t-1 th time node to the tth time node, wherein t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w;
determining a predicted state transition matrix Q 'by't:
When t is 2, Q't=Qt;
B, when t is more than 2 and less than or equal to m, Q'tElement q 'of (1)'ijt(wherein i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w) is obtained by an exponential smoothing method, namely:
q’ijt=αqijt+(1-α)q’ij(t-1)
wherein alpha is more than 0 and less than 1;
determining a predicted state transition matrix Q 'from t-1 time node to t time node'tThe algorithm of (1) inputs: actual state transition matrix QtParameter α, output: prediction state transition matrix Q't;
Wherein, the value t is sequentially 2, 3, … … and m, and all prediction state transition matrixes can be obtained;
3) obtaining an actual probability matrix
Let PtRepresenting the actual probability matrix at the t-th time node; note the book
Pt=[p1t p2t … pwt],pitThe ratio of the ith vulnerability number in the direct prediction set to all vulnerability numbers in the direct prediction set at the tth time node is represented, and the count is recordedWherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m, the method is executed, and the value of t is 1, 2, … … and m in sequence to obtain all actual probability matrixes;
4) obtaining a prediction probability matrix
Is P'tRepresenting a prediction probability matrix at the t-th time node; note the bookp’itThe predicted value of the ratio of the ith vulnerability in the direct prediction set to all vulnerabilities in the direct prediction set at the tth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m,
determining a prediction probability matrix by:
a is P 'when t is 1't=Pt;
B, when t is more than 1 and less than or equal to m, P'tOf (1) element p'itWherein i is more than or equal to 1 and less than or equal to w, and the value is obtained by an exponential smoothing method, namely: p'it=αpit+(1-α)p’i(t-1)Wherein alpha is more than 0 and less than 1,
determining a predicted probability matrix P 'of a t-th time node'tThe algorithm of (1) inputs: actual probability matrix PtParameter α, output: prediction state transition matrix P't;
Wherein, the value t is 1, 2, … … and m in sequence, and all prediction probability matrixes can be obtained;
5) obtaining the predicted value of directly predicting the number of various centralized bugs
Setting the actual value of the total number of the loopholes of the direct prediction set at each time node as C, and recording C as [ C ═ C%1 c2 … cmcm+1],ciRepresenting the total number of holes of the ith time node, i.e.Wherein i is more than or equal to 1 and less than or equal to m;
if the predicted value of the total number of vulnerabilities of the direct prediction set at each time node is C ', C ═ C'1 c’2 … c’mc’m+1]C 'is determined by'iWherein i is more than or equal to 1 and less than or equal to m:
(c 'when i is 1)'i=ci;
C 'when i is more than 1 and less than or equal to m + 1'iObtained by exponential smoothing, namely:
then c'm+1The prediction value of the total amount of the loopholes of the m +1 th time node of the direct prediction set is obtained;
obtaining a prediction state transition matrix Q 'from 2)'m(ii) a Obtaining a prediction probability matrix P 'from 4)'m;
According to the Markov algorithm: a matrix of the number proportion of all the vulnerabilities in the direct prediction set at the (m + 1) th time node of each vulnerability in the direct prediction set:
Pm+1=P’m·Q’m
according to the nature of the matrix multiplication, Pm+1Is a row vector containing w elements, Pm+1The ith element in (1) is pi(m+1)Wherein i is more than or equal to 1 and less than or equal to w;
setting the quantity prediction matrix of various loopholes in the (m + 1) th time node in the direct prediction set as R, and recording R ═ R1 r2… rw]Let us orderThen r isiThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to w;
the matrix R is the number prediction result of various vulnerabilities in the (m + 1) th time node in the direct prediction set S;
(3) predicting the quantity of various vulnerabilities in the indirect prediction set;
1) obtaining cosine similarity matrix
Set indirect prediction setIn v elements, i.e.All quantity information on m time nodes of the v-type vulnerability is contained; note the book The number of the ith vulnerability in the jth time node is represented, wherein w + v equals to n, i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to m;
definition ofThe variation vector of the ith vulnerability from the tth time node to the t +1 th time node isWherein i is more than or equal to 1 and less than or equal to v, and t is more than or equal to 1 and less than m;
defining the variation vector of the jth vulnerability from the tth time node to the t +1 th time node in the S asWherein j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m;
wherein the content of the first and second substances,andrespectively depict the change situation of the i-th vulnerability and the j-th vulnerability between two time nodes, and because the state transition has directionality,andis a variation vector;
therefore, from the t-th time node to the t + 1-th time node,the cosine similarity between the ith bug and the jth bug in S isWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than or equal to m;
is provided withThe cosine similarity between the ith vulnerability and the jth vulnerability in S is cos thetaijThe value is cos θijtWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m, namely:
set indirect prediction setThe cosine similarity matrix with the direct prediction set S is cos theta, thenWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
2) obtaining most similar vulnerabilities
Finding out the subscript j of the maximum value of the ith row in cos theta, wherein the jth vulnerability in the direct prediction set S is the indirect prediction setThe most similar loopholes of the ith loopholes, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setDirectly predicting the most similar vulnerabilities of various vulnerabilities in the set S;
3) obtaining the predicted value of indirectly predicting the number of various vulnerabilities in the set
Survey indirect prediction setDirectly predicting the most similar vulnerability in the set S, namely the jth vulnerability in the set S, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w; from the mth time node to the m +1 th time node, the relative increment of the jth vulnerability is
Then the predicted value of the number of the ith vulnerability at the (m + 1) th time nodeWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setPredicting the quantity of various loopholes at the (m + 1) th time node;
setting the quantity prediction matrix of various loopholes in the m +1 time node in the indirect prediction set asNote the bookOrder toThenThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to v;
matrix arrayI.e. indirect prediction setPredicting the number of the various loopholes in the (m + 1) th time node;
(4) obtaining a prediction result of a vulnerability corpus
Order toThen set RU=[R1 R2 … Rn]Namely the prediction result of the loophole complete set U at the m +1 time node, RiAnd the quantity predicted value of the ith type vulnerability in the vulnerability complete set U at the (m + 1) th time node is represented.
Further, the sum of each row in the state transition matrix in step 1) is 1, and the actual state transition matrix obtained by the above algorithm meets this requirement, which is proved as follows:
and (3) proving that:
(1) when s isit≥si(t-1)When there is qiit=1,qijt=0,j≠i;
(2) When s isit≥si(t-1)When there is qijt=fij·(1-qiit),j≠i;
further, the sum of each row in the state transition matrix in step 2) is 1, and the predicted state transition matrix obtained by the above algorithm meets this requirement, which is proved as follows:
and (3) proving that:
(1) when t is 2, there is Q't=Qt
So q'ijt=qijt,1≤i≤w,1≤j≤w;
(2) When t is more than 2 and less than or equal to m, there is q'ijt=αqijt+(1-α)q’ij(t-1)
According to the mathematical induction method:
2) When t is equal to k +1,
obtained from 1) and 2):when t is more than 2 and less than or equal to m, the method is established;
further, the probability matrix in step 3) is a row vector and the sum value is 1, and the actual probability matrix obtained by the above method meets the requirement, which proves as follows:
further, the probability matrix in step 4) is a row vector and the sum value is 1, and the prediction probability matrix obtained by the above algorithm meets the requirement, which proves as follows:
and (3) proving that:
(1) when t is 1, there is P't=Pt
So p'it=pit,1≤i≤w,1≤j≤w;
(2) When t is more than 1 and less than or equal to m, there is p'it=αpit+(1-α)p’i(t-1)
According to the mathematical induction method:
2) When t is equal to k +1,
further, the number of all kinds of security vulnerabilities in the authoritative information security vulnerability library is inspected and reported at a plurality of time nodes, and a vulnerability complete set is formed and expressed in a two-dimensional matrix form.
Further, the average quantity level of the loophole full set is inspected, the value is taken as the center, the step length is continuously increased, a proper neighborhood interval is determined, the interval is a direct prediction set, and the complement of the direct prediction set to the full set is an indirect prediction set.
The invention has the beneficial effects that:
and taking the security vulnerability as a research object, investigating historical data information of the security vulnerability to form a vulnerability complete set, and reasonably dividing the vulnerability complete set into a direct prediction set and an indirect prediction set. And then, predicting the direct prediction set by using a Markov method improved by an exponential smoothing method, establishing the relation between the direct prediction set and the indirect prediction set by using a cosine similarity principle, and further predicting the indirect prediction set. And finally, integrating the prediction results of the direct prediction set and the indirect prediction set, and providing a high-accuracy prediction value for workers in related fields.
Drawings
Fig. 1 is an algorithm diagram for dividing a vulnerability complete set into a direct prediction set and an indirect prediction set.
FIG. 2 is a graphical representation of a Markov prediction security vulnerability quantity method based on smoothing and similarity improvement.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
A Markov prediction vulnerability number method based on a smoothing method and similarity is shown in figures 1 and 2, and is characterized by comprising the following steps:
(1) firstly, the historical quantity information of the security vulnerability is inspected to form a vulnerability complete set, and the vulnerability complete set is divided into a direct prediction set and an indirect prediction set
Examining historical quantity information of the security vulnerabilities to form a vulnerability complete set, recording the vulnerability complete set as U, wherein the U comprises all quantity information of n types of vulnerabilities in m time nodes, and recording the quantity informationuijNumber of j time nodes representing i type vulnerabilityQuantity, wherein i is more than or equal to 1 and less than or equal to n, j is more than or equal to 1 and less than or equal to m, the average number level of the vulnerability complete set is considered, namely the arithmetic mean of all data in U is calculatedAnd rounding the calculated arithmetic mean down:
note that the i-th vulnerability is UiAnd (3) inspecting the average quantity level of various loopholes, namely calculating the arithmetic mean of data of each row in the U, and rounding down:
wherein i is more than or equal to 1 and less than or equal to n,
to be provided withTaking 1 as an initial step length as a reference, p as a parameter for determining the range of the direct prediction set, generally, p is more than 0 and less than 1, and in order to ensure the accuracy of the prediction result, the value range of p is recommended to be more than or equal to 0.5 and less than 1,
the algorithm is as follows: algorithm for dividing vulnerability complete set into direct prediction set and indirect prediction set
Inputting: vulnerability complete set U and parameter p
(2) Predicting the number of various vulnerabilities in the direct prediction set:
the method for predicting the number of various vulnerabilities in the direct prediction set mainly comprises the following steps:
1) obtaining an actual state transition matrix
Setting w elements in a direct prediction set S, namely S contains all quantity information of w types of vulnerabilities on m time nodes; note the booksijThe number of the ith vulnerability in the jth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to m,
let QtRepresenting the actual state transition matrix from the (t-1) th time node to the t-th time nodeqijtRepresenting the actual probability of transferring the ith vulnerability into the jth vulnerability from the (t-1) th time node to the tth time node, wherein t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w;
wherein d isijThe number of times that the number of the ith loopholes is reduced and the number of the jth loopholes is increased is represented, i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to w, and j is not equal to i; f. ofijIndicates the ratio of the number of i-th bugs decreasing and the number of j-th bugs increasing to the number of i-th bugs decreasing, i.e. the number of1≤i≤w,1≤j≤w,j≠i;qitRepresents QtWherein i is not less than 1 and not more than w;
the algorithm is as follows: determining qitActual state transition matrix Q from t-1 time node to t time nodetRow i q ofitIs calculated by
Inputting: direct prediction set S, parameter fij,
And (3) outputting: q. q.sit
Wherein, the value of i is 1, 2, … … and w in sequence, and a complete actual state transition matrix Q can be obtainedtTaking the t values of 2, 3, … … and m in sequence to obtain all actual state transition matrixes;
2) obtaining a predicted state transition matrix
Is Q'tRepresenting the predicted state transition matrix from the (t-1) th time node to the t-th time nodeq’ijtRepresenting the prediction probability of transferring the ith vulnerability into the jth vulnerability from the t-1 th time node to the tth time node, wherein t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w;
determining a predicted state transition matrix Q 'by't:
When t is 2, Q't=Qt;
B, when t is more than 2 and less than or equal to m, Q'tElement q 'of (1)'ijt(wherein i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w) is obtained by an exponential smoothing method, namely:
q’ijt=αqijt+(1-α)q’ij(t-1)
wherein alpha is more than 0 and less than 1;
determining a predicted state transition matrix Q 'from t-1 time node to t time node'tThe algorithm of (1) inputs: actual state transition matrix QtParameter α, output: prediction state transition matrix Q't;
Wherein, the value t is sequentially 2, 3, … … and m, and all prediction state transition matrixes can be obtained;
3) obtaining an actual probability matrix
Let PtRepresenting the actual probability matrix at the t-th time node; note Pt=[p1t p2t … pwt],pitThe ratio of the ith vulnerability number in the direct prediction set to all vulnerability numbers in the direct prediction set at the tth time node is represented, and the count is recordedWherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m, the method is executed, and the value of t is 1, 2, … … and m in sequence to obtain all actual probability matrixes;
4) obtaining a prediction probability matrix
Is P'tRepresenting a prediction probability matrix at the t-th time node; note the bookp’itThe predicted value of the ratio of the ith vulnerability in the direct prediction set to all vulnerabilities in the direct prediction set at the tth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m,
determining a prediction probability matrix by:
a is P 'when t is 1't=Pt;
B, when t is more than 1 and less than or equal to m, P'tOf (1) element p'itWherein i is more than or equal to 1 and less than or equal to w, and the value is obtained by an exponential smoothing method, namely: p'it=αpit+(1-α)p’i(t-1)Wherein alpha is more than 0 and less than 1,
determining a predicted probability matrix P 'of a t-th time node'tThe algorithm of (1) inputs: actual probability matrix PtParameter α, output: prediction state transition matrix P't;
Wherein, the value t is 1, 2, … … and m in sequence, and all prediction probability matrixes can be obtained;
5) obtaining the predicted value of directly predicting the number of various centralized bugs
Setting the actual value of the total number of the loopholes of the direct prediction set at each time node as C, and recording C as [ C ═ C%1 c2 … cmcm+1],ciRepresenting the total number of holes of the ith time node, i.e.Wherein i is more than or equal to 1 and less than or equal to m;
if the predicted value of the total number of vulnerabilities of the direct prediction set at each time node is C ', C ═ C'1 c’2 … c’mc’m+1]C 'is determined by'iWherein i is more than or equal to 1 and less than or equal to m:
(c 'when i is 1)'i=ci;
C 'when i is more than 1 and less than or equal to m + 1'iObtained by exponential smoothing, namely:
then c'm+1The prediction value of the total amount of the loopholes of the m +1 th time node of the direct prediction set is obtained;
obtaining a prediction state transition matrix Q 'from 2)'m(ii) a Obtaining a prediction probability matrix P 'from 4)'m;
According to the Markov algorithm: a matrix of the number proportion of all the vulnerabilities in the direct prediction set at the (m + 1) th time node of each vulnerability in the direct prediction set:
Pm+1=P’m·Q’m
according to the nature of the matrix multiplication, Pm+1Is aA row vector comprising w elements, denoted Pm+1The ith element in (1) is pi(m+1)Wherein i is more than or equal to 1 and less than or equal to w;
setting the quantity prediction matrix of various loopholes in the (m + 1) th time node in the direct prediction set as R, and recording R ═ R1 r2… rw]Let us orderThen r isiThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to w;
the matrix R is the number prediction result of various vulnerabilities in the (m + 1) th time node in the direct prediction set S;
(3) predicting the quantity of various vulnerabilities in the indirect prediction set;
1) obtaining cosine similarity matrix
Set indirect prediction setIn v elements, i.e.All quantity information on m time nodes of the v-type vulnerability is contained; note the book The number of the ith vulnerability in the jth time node is represented, wherein w + v equals to n, i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to m;
definition ofThe variation vector of the ith vulnerability from the tth time node to the t +1 th time node isWherein i is more than or equal to 1 and less than or equal to v, and t is more than or equal to 1 and less than m;
defining class j leaks in SThe change vector of the hole from the t-th time node to the t + 1-th time node isWherein j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m;
wherein the content of the first and second substances,andrespectively depict the change situation of the i-th vulnerability and the j-th vulnerability between two time nodes, and because the state transition has directionality,andis a variation vector;
therefore, from the t-th time node to the t + 1-th time node,the cosine similarity between the ith bug and the jth bug in S isWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than or equal to m;
is provided withThe cosine similarity between the ith vulnerability and the jth vulnerability in S is cos thetaijThe value is cos θijtWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m, namely:
set indirect prediction setThe cosine similarity matrix with the direct prediction set S is cos theta, thenWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
2) obtaining most similar vulnerabilities
Finding out the subscript j of the maximum value of the ith row in cos theta, wherein the jth vulnerability in the direct prediction set S is the indirect prediction setThe most similar loopholes of the ith loopholes, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setDirectly predicting the most similar vulnerabilities of various vulnerabilities in the set S;
3) obtaining the predicted value of indirectly predicting the number of various vulnerabilities in the set
Survey indirect prediction setDirectly predicting the most similar vulnerability in the set S, namely the jth vulnerability in the set S, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w; from the mth time node to the m +1 th time node, the relative increment of the jth vulnerability is
Then the predicted value of the number of the ith vulnerability at the (m + 1) th time nodeWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setPredicting the quantity of various loopholes at the (m + 1) th time node;
setting the quantity prediction matrix of various loopholes in the m +1 time node in the indirect prediction set asNote the bookOrder toThenThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to v;
matrix arrayI.e. indirect prediction setPredicting the number of the various loopholes in the (m + 1) th time node;
(4) obtaining a prediction result of a vulnerability corpus
Order toThen set RU=[R1 R2 … Rn]Namely the prediction result of the loophole complete set U at the m +1 time node, RiAnd the quantity predicted value of the ith type vulnerability in the vulnerability complete set U at the (m + 1) th time node is represented.
Further, the sum of each row in the state transition matrix in step 1) is 1, and the actual state transition matrix obtained by the above algorithm meets this requirement, which is proved as follows:
and (3) proving that:
(3) when s isit≥si(t-1)When there is qiit=1,qijt=0,j≠i;
(4) When s isit≥si(t-1)When there is qijt=fij·(1-qiit),j≠i;
further, the sum of each row in the state transition matrix in step 2) is 1, and the predicted state transition matrix obtained by the above algorithm meets this requirement, which is proved as follows:
and (3) proving that:
(1) when t is 2, there is Q't=Qt
So q'ijt=qijt,1≤i≤w,1≤j≤w;
(2) When t is more than 2 and less than or equal to m, there is q'ijt=αqijt+(1-α)q’ij(t-1)
According to the mathematical induction method:
4) When t is equal to k +1,
obtained from 1) and 2):when t is more than 2 and less than or equal to m, the method is established;
further, the probability matrix in step 3) is a row vector and the sum value is 1, and the actual probability matrix obtained by the above method meets the requirement, which proves as follows:
further, the probability matrix in step 4) is a row vector and the sum value is 1, and the prediction probability matrix obtained by the above algorithm meets the requirement, which proves as follows:
and (3) proving that:
(1) when t is 1, there is P't=Pt
So p isit=pit,1≤i≤w,1≤j≤w;
(2) When t is more than 1 and less than or equal to m, there is p'it=αpit+(1-α)p’i(t-1)
According to the mathematical induction method:
2) When t is equal to k +1,
further, the number of all kinds of security vulnerabilities in the authoritative information security vulnerability library is inspected and reported at a plurality of time nodes, and a vulnerability complete set is formed and expressed in a two-dimensional matrix form.
Further, the average quantity level of the loophole full set is inspected, the value is taken as the center, the step length is continuously increased, a proper neighborhood interval is determined, the interval is a direct prediction set, and the complement of the direct prediction set to the full set is an indirect prediction set.
Claims (7)
1. A Markov prediction vulnerability quantity method based on a smoothing method and similarity is characterized by comprising the following steps:
(1) firstly, the historical quantity information of the security vulnerability is inspected to form a vulnerability complete set, and the vulnerability complete set is divided into a direct prediction set and an indirect prediction set
Examining historical quantity information of the security vulnerabilities to form a vulnerability complete set, recording the vulnerability complete set as U, wherein the U comprises all quantity information of n types of vulnerabilities in m time nodes, and recording the quantity informationuijRepresenting the number of the ith vulnerability in the jth time node, wherein i is more than or equal to 1 and less than or equal to n, j is more than or equal to 1 and less than or equal to m, and inspecting the average number level of the vulnerability complete set, namely calculating the arithmetic mean of all data in UAnd rounding the calculated arithmetic mean down:
note that the i-th vulnerability is UiAnd (3) inspecting the average quantity level of various loopholes, namely calculating the arithmetic mean of data of each row in the U, and rounding down:
wherein i is more than or equal to 1 and less than or equal to n,
to be provided withFor reference, 1 is the initial step size, p is for determining the range of the direct prediction setThe parameters are generally more than 0 and less than 1, in order to ensure the accuracy of the prediction result,
the algorithm is as follows: algorithm for dividing vulnerability complete set into direct prediction set and indirect prediction set
Inputting: vulnerability complete set U and parameter p
(2) Predicting the number of various vulnerabilities in the direct prediction set:
the method for predicting the number of various vulnerabilities in the direct prediction set mainly comprises the following steps:
1) obtaining an actual state transition matrix
Setting w elements in a direct prediction set S, namely S contains all quantity information of w types of vulnerabilities on m time nodes; note the booksijThe number of the ith vulnerability in the jth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to m,
let QtRepresenting the actual state transition from the t-1 st time node to the t-th time nodeqijtIndicating that the ith vulnerability is transferred to the jth vulnerability from the t-1 th time node to the tth time nodeThe actual probability of t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and j is more than or equal to 1 and less than or equal to w;
wherein d isijThe number of times that the number of the ith loopholes is reduced and the number of the jth loopholes is increased is represented, i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to w, and j is not equal to i; f. ofijIndicates the ratio of the number of i-th bugs decreasing and the number of j-th bugs increasing to the number of i-th bugs decreasing, i.e. the number ofqitRepresents QtWherein i is not less than 1 and not more than w;
the algorithm is as follows: determining qitActual state transition matrix Q from t-1 time node to t time nodetRow i q ofitIs calculated by
Inputting: direct prediction set S, parameter fij,
And (3) outputting: q. q.sit
Wherein, the value of i is 1, 2, … … and w in sequence, and a complete actual state transition matrix Q can be obtainedtTaking the t values of 2, 3, … … and m in sequence to obtain all actual state transition matrixes;
2) obtaining a predicted state transition matrix
Is Q'tRepresenting the predicted state transition matrix from the t-1 st time node to the t-th time nodeq′ijtThe prediction probability of the ith vulnerability transferred from the t-1 th time node to the jth vulnerability is represented, wherein t is more than 1 and less than or equal to m, i is more than or equal to 1 and less than or equal to w, and 1 is more than or equal to 1j≤w;
Determining a predicted state transition matrix Q 'by't:
When t is 2, Q't=Qt;
B, when t is more than 2 and less than or equal to m, Q'tElement q 'of (1)'ijtWherein i is more than or equal to 1 and less than or equal to w, j is more than or equal to 1 and less than or equal to w is obtained by an exponential smoothing method, namely:
q′ijt=αqijt+(1-α)q′ij(t-1)
wherein alpha is more than 0 and less than 1;
determining a predicted state transition matrix Q 'from t-1 time node to t time node'tThe algorithm of (1) inputs: actual state transition matrix QtParameter α, output: prediction state transition matrix Q't;
Wherein, the value t is sequentially 2, 3, … … and m, and all prediction state transition matrixes can be obtained;
3) obtaining an actual probability matrix
Let PtRepresenting the actual probability matrix at the t-th time node; note Pt=[p1t p2t…pwt],pitThe ratio of the ith vulnerability number in the direct prediction set to all vulnerability numbers in the direct prediction set at the tth time node is represented, and the count is recordedWherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m, the method is executed, and the value of t is 1, 2, … … and m in sequence to obtain all actual probability matrixes;
4) obtaining a prediction probability matrix
Is P'tRepresenting a prediction probability matrix at the t-th time node; note the bookp′itThe predicted value of the ratio of the ith vulnerability in the direct prediction set to all vulnerabilities in the direct prediction set at the tth time node is represented, wherein i is more than or equal to 1 and less than or equal to w, t is more than or equal to 1 and less than or equal to m,
determining a prediction probability matrix by:
a is P 'when t is 1't=Pt;
B, when t is more than 1 and less than or equal to m, P'tOf (1) element p'itWherein i is more than or equal to 1 and less than or equal to w, and the value is obtained by an exponential smoothing method, namely: p'it=αpit+(1-α)p′i(t-1)Wherein alpha is more than 0 and less than 1,
determining a predicted probability matrix P 'of a t-th time node'tThe algorithm of (1) inputs: actual probability matrix PtParameter α, output: prediction state transition matrix P't;
Wherein, the value t is 1, 2, … … and m in sequence, and all prediction probability matrixes can be obtained;
5) obtaining the predicted value of directly predicting the number of various centralized bugs
Setting the actual value of the total number of the loopholes of the direct prediction set at each time node as C, and recording C as [ C ═ C%1 c2…cm cm+1],ciRepresenting the total number of holes of the ith time node, i.e.Wherein i is more than or equal to 1 and less than or equal to m;
if the predicted value of the total number of vulnerabilities of the direct prediction set at each time node is C ', C ═ C'1 c′2…c′m c′m+1]C 'is determined by'iWherein i is more than or equal to 1 and less than or equal to m:
(c 'when i is 1)'i=ci;
C 'when i is more than 1 and less than or equal to m + 1'iObtained by exponential smoothing, i.e.:
then c'm+1The prediction value of the total amount of the loopholes of the m +1 th time node of the direct prediction set is obtained;
obtaining a prediction state transition matrix Q 'from 2)'m(ii) a Obtaining a prediction probability matrix P 'from 4)'m;
According to the Markov algorithm: a matrix of the number proportion of all the vulnerabilities in the direct prediction set at the (m + 1) th time node of each vulnerability in the direct prediction set:
Pm+1=P′m·Q′m
according to the nature of the matrix multiplication, Pm+1Is a row vector containing w elements, Pm+1The ith element in (1) is pi(m+1)Wherein i is more than or equal to 1 and less than or equal to w;
setting the quantity prediction matrix of various loopholes in the (m + 1) th time node in the direct prediction set as R, and recording R ═ R1 r2…rw]Let us orderThen r isiThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to w;
the matrix R is the number prediction result of various vulnerabilities in the (m + 1) th time node in the direct prediction set S;
(3) predicting the quantity of various vulnerabilities in the indirect prediction set;
1) obtaining cosine similarity matrix
Set indirect prediction setIn v elements, i.e.All quantity information on m time nodes of the v-type vulnerability is contained;note the book The number of the ith vulnerability in the jth time node is represented, wherein w + v equals to n, i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to m;
definition ofThe variation vector of the ith vulnerability from the tth time node to the t +1 th time node isWherein i is more than or equal to 1 and less than or equal to v, and t is more than or equal to 1 and less than m;
defining the variation vector of the jth vulnerability from the tth time node to the t +1 th time node in the S asWherein j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m;
wherein the content of the first and second substances,andrespectively depict the change situation of the i-th vulnerability and the j-th vulnerability between two time nodes, and because the state transition has directionality,andis a variation vector;
therefore, from the t-th time node to the t + 1-th time node,the cosine similarity between the ith bug and the jth bug in S isWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than or equal to m;
is provided withThe cosine similarity between the ith vulnerability and the jth vulnerability in S is cos thetaijThe value is cos θijtWherein i is more than or equal to 1 and less than or equal to v, j is more than or equal to 1 and less than or equal to w, and t is more than or equal to 1 and less than m, namely:
set indirect prediction setThe cosine similarity matrix with the direct prediction set S is cos theta, thenWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
2) obtaining most similar vulnerabilities
Finding out the subscript j of the maximum value of the ith row in cos theta, wherein the jth vulnerability in the direct prediction set S is the indirect prediction setThe most similar loopholes of the ith loopholes, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setIn the direct prediction set SThe most similar vulnerability of (1);
3) obtaining the predicted value of indirectly predicting the number of various vulnerabilities in the set
Survey indirect prediction setDirectly predicting the most similar vulnerability in the set S, namely the jth vulnerability in the set S, wherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w; from the mth time node to the m +1 th time node, the relative increment of the jth vulnerability is
Then the predicted value of the number of the ith vulnerability at the (m + 1) th time nodeWherein i is more than or equal to 1 and less than or equal to v, and j is more than or equal to 1 and less than or equal to w;
executing the above operations, and sequentially taking 1, 2, … … and v as the value of i to obtain an indirect prediction setPredicting the quantity of various loopholes at the (m + 1) th time node;
setting the quantity prediction matrix of various loopholes in the m +1 time node in the indirect prediction set asNote the bookOrder toThenThe number predicted value of the ith vulnerability at the (m + 1) th time node is represented, wherein i is more than or equal to 1 and is less than or equal to v;
matrix arrayI.e. indirect prediction setPredicting the number of the various loopholes in the (m + 1) th time node;
(4) obtaining a prediction result of a vulnerability corpus
2. The method according to claim 1, wherein the sum of each row in the state transition matrix in step 1) is 1, and the actual state transition matrix obtained by the method for predicting the number of the holes based on the markov algorithm and the similarity satisfies the requirement as follows:
and (3) proving that:
(1) when s isit≥si(t-1)When q is greateriit=1,qijt=0,j≠i;
(2) When s isit≥si(t-1)When q is greaterijt=fij·(1-qiit),j≠i;
Therefore, it is
3. the method for predicting the number of the holes based on the markov algorithm with the similarity as claimed in claim 1, wherein the sum of each row in the state transition matrix in the step 2) is 1, and the predicted state transition matrix obtained by the method for predicting the number of the holes based on the markov algorithm with the similarity meets the requirement, which is proved as follows:
and (3) proving that:
(1) when t is 2, there is Q't=Qt
So q'ijt=qijt,1≤i≤w,1≤j≤w;
(2) When t is more than 2 and less than or equal to m, there is q'ijt=αqijt+(1-α)q′ij(t-1)
According to the mathematical induction method:
2) When t is equal to k +1,
obtained from 1) and 2):when t is more than 2 and less than or equal to m, the method is established;
4. the method according to claim 1, wherein the probability matrix in step 3) is a row vector and the sum is 1, and the actual probability matrix obtained by the method for predicting the number of the holes based on the markov algorithm and the similarity satisfies the requirement as follows:
5. the method for predicting the number of the holes based on the Markov algorithm with the smoothness and the similarity as claimed in claim 1, wherein the probability matrix in the step 4) is a row vector and the sum is 1, and the prediction probability matrix obtained by the method for predicting the number of the holes based on the Markov algorithm with the similarity meets the requirement, which is proved as follows:
and (3) proving that:
(1) when t is 1, there is P't=Pt
So p'it=pit,1≤i≤w,1≤j≤w;
(2) When t is more than 1 and less than or equal to m, there is p'it=αpit+(1-α)p′i(t-1)
According to the mathematical induction method:
2) When t is equal to k +1,
6. the method for predicting the number of the vulnerabilities based on the markov algorithm with the similarity as claimed in claim 1, wherein the number of the various types of vulnerabilities in the authority information security vulnerability database at a plurality of time nodes is examined to form a vulnerability complete set, which is expressed in a two-dimensional matrix form.
7. The method of claim 1, wherein an average number level of a vulnerability corpus is examined, the step length is continuously increased by taking the average number as a center, a proper neighborhood interval is determined, the interval is a direct prediction set, and a complement of the direct prediction set to the corpus is an indirect prediction set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810701155.6A CN108959084B (en) | 2018-06-29 | 2018-06-29 | Markov vulnerability prediction quantity method based on smoothing method and similarity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810701155.6A CN108959084B (en) | 2018-06-29 | 2018-06-29 | Markov vulnerability prediction quantity method based on smoothing method and similarity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108959084A CN108959084A (en) | 2018-12-07 |
CN108959084B true CN108959084B (en) | 2022-03-25 |
Family
ID=64484081
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810701155.6A Active CN108959084B (en) | 2018-06-29 | 2018-06-29 | Markov vulnerability prediction quantity method based on smoothing method and similarity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108959084B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112115476B (en) * | 2020-08-06 | 2023-10-24 | 扬州大学 | Automatic vulnerability classification method, system and computer equipment based on LSTM |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101923618A (en) * | 2010-08-19 | 2010-12-22 | 中国航天科技集团公司第七一○研究所 | Hidden Markov model based method for detecting assembler instruction level vulnerability |
CN103532761A (en) * | 2013-10-18 | 2014-01-22 | 嘉兴学院 | Survivability evaluating method applicable to attacked wireless sensing network |
CN104469798A (en) * | 2014-12-12 | 2015-03-25 | 重庆邮电大学 | Communication network load condition information forecasting method based on Markov chain |
EP3139318A1 (en) * | 2015-09-04 | 2017-03-08 | Siemens Aktiengesellschaft | Patch management for industrial control systems |
-
2018
- 2018-06-29 CN CN201810701155.6A patent/CN108959084B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101923618A (en) * | 2010-08-19 | 2010-12-22 | 中国航天科技集团公司第七一○研究所 | Hidden Markov model based method for detecting assembler instruction level vulnerability |
CN103532761A (en) * | 2013-10-18 | 2014-01-22 | 嘉兴学院 | Survivability evaluating method applicable to attacked wireless sensing network |
CN104469798A (en) * | 2014-12-12 | 2015-03-25 | 重庆邮电大学 | Communication network load condition information forecasting method based on Markov chain |
EP3139318A1 (en) * | 2015-09-04 | 2017-03-08 | Siemens Aktiengesellschaft | Patch management for industrial control systems |
US10331429B2 (en) * | 2015-09-04 | 2019-06-25 | Siemens Aktiengesellschaft | Patch management for industrial control systems |
Non-Patent Citations (1)
Title |
---|
基于漏洞严重程度分类的漏洞预测模型;高志伟等;《电子学报》;20130930;第41卷(第9期);第1784-1787页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108959084A (en) | 2018-12-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Saylor et al. | Quantifying comparison of large detrital geochronology data sets | |
CN108647272B (en) | Method for predicting concentration of butane at bottom of debutanizer by expanding small samples based on data distribution | |
CN113038302B (en) | Flow prediction method and device and computer storage medium | |
Yang et al. | Expected efficiency based on directional distance function in data envelopment analysis | |
Manninen et al. | Leukemia prediction using sparse logistic regression | |
US7587280B2 (en) | Genomic data mining using clustering logic and filtering criteria | |
Villoria et al. | Gaussian quadratures vs. Monte Carlo experiments for systematic sensitivity analysis of computable general equilibrium model results | |
CN115510981A (en) | Decision tree model feature importance calculation method and device and storage medium | |
WO2023201772A1 (en) | Cross-domain remote sensing image semantic segmentation method based on adaptation and self-training in iteration domain | |
CN114611582B (en) | Method and system for analyzing substance concentration based on near infrared spectrum technology | |
CN108959084B (en) | Markov vulnerability prediction quantity method based on smoothing method and similarity | |
CN110324178B (en) | Network intrusion detection method based on multi-experience nuclear learning | |
Weine et al. | Application of equal local levels to improve QQ plot testing bands with R package qqconf | |
Saidov | Data visualization and its proof by compactness criterion of objects of classes | |
Salem et al. | Sequential dimension reduction for learning features of expensive black-box functions | |
Zhang et al. | Dbiecm-an evolving clustering method for streaming data clustering | |
CN114298659A (en) | Data processing method and device for evaluation object index and computer equipment | |
Izsák | Some practical aspects of fitting and testing the Zipf-Mandelbrot model: A short essay | |
Liu et al. | A robust regression based on weighted LSSVM and penalized trimmed squares | |
Zhao et al. | Distribution-free and model-free multivariate feature screening via multivariate rank distance correlation | |
Mazzocco et al. | Estimates and impact of lymphocyte division parameters from CFSE data using mathematical modelling | |
Bongiorno et al. | Statistically validated hierarchical clustering: Nested partitions in hierarchical trees | |
RU2586025C2 (en) | Method for automatic clustering of objects | |
Zhang et al. | Predicting Future Event via Small Data (eg, 4 Data) by ASF and Curve Fitting Methods | |
Ren et al. | Development of feature extraction method based on interval-valued Pythagorean fuzzy decision theory |
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 |