CN102467635A - Prediction method for trojan horse - Google Patents
Prediction method for trojan horse Download PDFInfo
- Publication number
- CN102467635A CN102467635A CN2011101832196A CN201110183219A CN102467635A CN 102467635 A CN102467635 A CN 102467635A CN 2011101832196 A CN2011101832196 A CN 2011101832196A CN 201110183219 A CN201110183219 A CN 201110183219A CN 102467635 A CN102467635 A CN 102467635A
- Authority
- CN
- China
- Prior art keywords
- wooden horse
- colony
- binary tree
- prediction method
- collection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a prediction method for the trojan horse based on genetic algorithm of binary tree modeling, which can search suitable model structures automatically. During the prediction process, performances can be evaluated by assessing the difference between the predicted trojan event amount and the actual trojan event amount by using a target function value; the evaluated amount and the actual amount can be more close to each other by adjusting the target function to finally achieve good approaching to the prediction function so as to achieve a better performance index and satisfy predicted requirements.
Description
Technical field
The invention belongs to network security, relate in particular to wooden horse event prediction method.
Background technology
Current, along with the high speed development of infotech, the network size of internet, network information amount and network application etc. are all in continuous growth.The internet is in the each side field that relates to people life, like politics, and commerce, finance, culture and education, more and more important effect is being brought into play in communication etc.But the internet also is faced with increasing network safety event bringing people simultaneously greatly easily.
The network safety event forecasting techniques is fully to collect the current network flow situation, obtaining a special kind of skill of predicting on the historical security incident of the network basis that a situation arises.Because wooden horse is attacked and in the security incident under the large scale network, is accounted for significant proportion; Through wooden horse incident generation quantity in the moment in the future network is predicted; Can make things convenient for the network management personnel that the cardinal principle situation of whole network is had a preliminary judgement on the one hand; And formulate the network security policy that conforms to it according to situation about judging; Can shift to an earlier date the network disaster that anticipation will take place on the other hand or attack, and in time take counter-measure with attacking before taking place, eliminate problem in bud in disaster.
Existing wooden horse event prediction method has following several kinds:
Linear regression method: the autoregressive moving-average model with classics is representative; These class methods think that following wooden horse incident generation quantity is to receive directly influencing of former quantity and noise; Therefore predicted value promptly is the weighted sum of historical data and noise data, and expression formula is following:
Wherein p is the autoregressive model exponent number, and q is the moving average model exponent number, and x is historical observation data,
(i=1,2, L, p), θ
j(j=1,2, L q) is respectively auto-regressive parameter and running mean parameter.Be characterized in that model is simple, realize easily, but it need the parameter of user's proper configuration model on the one hand; This needs the user to possess corresponding field experience; Limited the use of algorithm, not accurate enough to this method on the other hand to approaching of anticipation function, so prediction effect is good inadequately.
The method of rule-based discovery: like the sequential rule discovery, these class methods are through according to frequent item set sequencing in time, release their sequential correlation rule, and according to this rule the time that the back will take place are predicted.This method is not owing to setting up a clear structure model to predicting this complex nonlinear problem, so prediction effect is good inadequately.
Summary of the invention
Therefore, the objective of the invention is to problem, provide a kind of Forecasting Methodology to satisfy the prediction accuracy of network security to wooden horse incident generation quantity, the requirement of aspects such as time complexity based on the binary tree modeling to wooden horse incident generation quantitative forecast in the network.
The objective of the invention is to realize through following technical scheme:
The invention provides a kind of wooden horse event prediction method, may further comprise the steps:
Step 1) is confirmed termination set and collection of functions, and said termination set comprises variable and constant, and the element in the said collection of functions then is to be used for operational symbol that the element in the termination set is operated;
Step 2) generate a series of function expressions at random according to termination set and collection of functions, said function expression is used to calculate current wooden horse incident generation quantity, and it is input as the quantity that the wooden horse incident takes place in the historical time section;
Step 3) is represented each function expression that the root node of said binary tree is in the collection of functions picked at random with the form of binary tree, intermediate node can be selected in collection of functions and termination set at random, and leaf node is selected in termination set at random;
Step 4) with whole binary tree colony as initial parent colony, to come execution genetic algorithm as evaluation to said initial parent colony to fitness according to the wooden horse quantity in the current network that said function expression was calculated and the difference of actual wooden horse quantity;
Step 5) utilizes the minimum function expression of the pairing said difference of the final optimized individual that generates of step 4) to come the generation quantity of the wooden horse incident in future in the network is predicted.
According to the wooden horse event prediction method of the embodiment of the invention, wherein, also comprise the step that termination set and collection of functions are encoded before the step 3): each the element in termination set and the collection of functions is encoded with natural number.
According to the wooden horse event prediction method of the embodiment of the invention, wherein, the depth capacity of every tree gets 4～6 in the step 3).
According to the wooden horse event prediction method of the embodiment of the invention, wherein step 4) may further comprise the steps:
Step 4-1) parent colony is carried out fitness evaluation, from parent colony, select to carry out the progeny population of genetic manipulation;
Step 4-2) to step 4-1) progeny population that obtains carries out interlace operation and mutation operation;
Step 4-3) with step 4-2) the binary tree colony that obtains is as parent colony, repeated execution of steps 4-1), 4-2) and 4-3), till the multiplicity that satisfies appointment.
According to the wooden horse event prediction method of the embodiment of the invention, wherein step 4-1) may further comprise the steps: adopt the method for roulette to come from parent colony, to select the high binary tree colony of fitness; Every in choosing binary tree colony tree is carried out replicate run; With the binary tree colony of being duplicated as progeny population.
According to the wooden horse event prediction method of the embodiment of the invention, wherein, step 4-2) crossover probability is 0.9 in; The variation probability is 0.05.Step 4-3) multiplicity described in is 100 times.
Existing Forecasting Methodology all need be set up forecast model; But be difficult to obtain the formula that embodies of anticipation function again; And the wooden horse event prediction method of genetic algorithm based on the binary tree modeling proposed by the invention can be searched for the proper model structure automatically, in forecasting process through using a target function value to come the wooden horse incident quantity that evaluation prediction comes out and the difference of actual wooden horse incident quantity to come assess performance; And, make that assessment quantity and actual quantity are more and more approaching through constantly adjusting this objective function, and finally reach good approximation, thereby reached the preferable performance index anticipation function, satisfied the demand of expection.
Description of drawings
Followingly the embodiment of the invention is described further with reference to accompanying drawing, wherein:
Fig. 1 is according to an embodiment of the invention based on the wooden horse event prediction method flow diagram of binary tree modeling genetic algorithm;
Fig. 2 is according to the binary tree synoptic diagram behind the function expression coding of the embodiment of the invention;
Fig. 3 is the binary tree parent colony synoptic diagram that will carry out heredity according to the embodiment of the invention;
Fig. 4 is the progeny population synoptic diagram that obtains behind the execution selection operation according to the embodiment of the invention;
Fig. 5 carries out the binary tree synoptic diagram that obtains after the interlace operation to progeny population shown in Figure 4;
Fig. 6 carries out the binary tree synoptic diagram that obtains behind the mutation operation to progeny population shown in Figure 5.
Embodiment
In order to make the object of the invention, technical scheme and advantage are clearer, pass through specific embodiment to further explain of the present invention below in conjunction with accompanying drawing.Should be appreciated that specific embodiment described herein only in order to explanation the present invention, and be not used in qualification the present invention.
In order to understand the present invention better, carry out brief account to genetic algorithm with based on the principle of binary tree modeling genetic algorithm earlier.The step that genetic algorithm generally comprises is: 1) select an initial population; 2) estimate each individual fitness; 3) carry out selection operation; 4) repeating step 5)-7) until satisfying termination condition; 5) intersect and mutation operation; 6) estimate each individual fitness, 7) carry out selection operation.Based on binary tree modeling genetic algorithms use the thought of genetic algorithm, through to individual constantly duplicating in the population, intersect and mutation operation, search for whole solution space, to obtain optimum solution.The difference of itself and general genetic algorithm is in the processing of initial population, produces a series of binary trees at random, and the nodal value of this tree is taken from two set, i.e. termination set and collection of functions.Element in the termination set is to want operated numerical value, and the element in the collection of functions then is to be used for the operational symbol that logarithm value operates.And the object of genetic algorithm operation is exactly a whole binary tree colony.For example replicate run then is to choose a binary tree, to its whole duplicating.Interlace operation then is to choose two nodes in the binary tree, and the position that faces toward two nodes exchanges.Mutation operation then is a node of choosing in the binary tree, is the former affiliated binary tree of sub-binary tree disengaging of root node with this node, forms an independently binary tree.
The process flow diagram of an embodiment of the wooden horse event prediction method that is based on binary tree modeling genetic algorithm shown in Figure 1.In order to make wooden horse event prediction method satisfy user's demand, adopt in an embodiment of the present invention based on the thought of binary tree modeling genetic algorithm and predict.The first step is to establish an objective function, and this objective function will make the difference of wooden horse quantity in the current network that calculates and actual wooden horse quantity minimum, and this difference also can be called fitness evaluation.Algorithm is encoded to termination set and collection of functions element then; The purpose of encoding here is in order in binary tree, can better to represent; And utilize these two set initialization binary trees to gather; Begin evolutionary process afterwards, the parent of in evolutionary process each time, selecting to have superperformance carries out genetic manipulation, finally reach end condition then algorithm stop.
As shown in Figure 1, at first confirm objective function y=F (x
1, x
2..., x
k), k independent variable and a dependent variable are arranged, then independent variable is the quantity X=(x that the wooden horse incident takes place in the historical time section
1, x
2..., x
k), dependent variable y representes that the current wooden horse incident generation quantity that calculates, y ' they are wooden horse quantity actual in the current network, confirm an optimal function expression formula F (x
1, x
2..., x
k), make the difference ε of wooden horse quantity in the current network that calculates and actual wooden horse quantity minimum, promptly have: min ε=| F (x
1, x
2..., x
k)-y ' | set up, above-mentioned expression formula is also referred to as fitness function.F (x wherein
1, x
2..., x
k) be the k meta-function, x
iBe i independent variable of function, i=1,2 ..., k.Difference ε also can be called fitness evaluation.Difference ε is more little, represents fitness high more, otherwise difference ε is big more, represents fitness more little.The expression formula of concrete objective function F generates in the initialized stage at random; But Once you begin after the operation of genetic algorithm; Carry out genetic manipulation according to the size of fitness value exactly, keep the big individuality of fitness at last, eliminate the little individuality of fitness.
The nodal value of the binary tree that will generate is taken from the element in these two set.The element of termination set T is decided to be variable x and constant c.The element of collection of functions W be decided to be arithmetic+,-, * ,/and elementary function computing { sin, tan, log}.These functional operation can enchashment be had any functional operation symbol in the technology, but generally got arithmetic and elementary function computing by artificial appointment, because can in solution space, search for effectively like this, can not skip optimum solution again.In order in program, can better to represent and computing, can also encode to the element in these two set.Concrete encoding scheme is as shown in table 1:
Table 1
| Element in the set | c | x | + | - | * | / | Sin | Tan | log |
| Encoded radio | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
Wherein encoded radio is 0 to show it is constant c, on interval [0,1], gets a random number accordingly, for example can come to get arbitrarily a random number with a random number generator and get final product.Encoded radio is 1 to show and take from variable x
i(i=1,2, L L L, some in k), then objective function expression formula F (x
1, x
2, L L, x
k) can express with the mode behind the coding.Hypothetical target function F (x
1, x
2, x
3, x
4)=x
1* x
2+ x
3* x
4, then the binary tree expression-form behind the coding of this function expression is as shown in Figure 2.Should understand this expression formula is from collection of functions and termination set, to choose element at random to generate, and does not have practical significance, only is for the binary tree expression-form in the key diagram 2.
Then, after coding, to utilize these two set to come initialization parent colony.Parent colony promptly is each selected object colony that carries out genetic manipulation with respect to progeny population, and the colony that generates after genetic manipulation finishes is called progeny population, and this progeny population promptly is the parent colony that carries out genetic manipulation next time.The genetic manipulation concrete steps are to produce 100 binary trees, that is to say 100 objective functions.Wherein the depth capacity of every tree (number of plies) gets 4～6, generally gets in the prior art in 10, and major part is all got 4～6 and got final product, so as to algorithm search to the optimal function expression formula carry out analysis interpretation.The fundamental element of tree makes up as follows: when producing 100 trees of initial population; The root node of every tree is picked at random in collection of functions W corresponding codes value; Intermediate node can be selected in collection of functions W and termination set T corresponding codes value at random, and leaf node is selected in termination set T corresponding codes value at random.In order to make whole binary tree colony can carry out suitable genetic evolution, guarantee that again the complexity of system-computed is unlikely too high simultaneously, it is 100 that population size is set here.Normally population size get any value all can, but to consider problems such as machine performance, the time complexity of computing, general population scale is got 100 and is got final product.Make up the operand of whole genetic algorithm through above-mentioned steps.
Then, carry out fitness evaluation, select to want the object of genetic manipulation according to fitness.
Each individuality of representing with the binary tree form in the parent colony decoded according to the corresponding relation of table 1 draws individual corresponding function expression F, and F bring into fitness function min ε=| F (x
1, x
2..., x
k)-y ' |, thus corresponding 100 fitness values, i.e. f (i), (i=1,2, L L L, 100), this value is more little, shows that then this individual fitness is high more.Suppose that the value of each independent variable is here: x
1=10, x
2=5, x
3=20, x
4=2, the target function value that then calculates is 90, and the value of actual wooden horse quantity is y '=80 in the network, and the error that then obtains this moment is ε=10.This value is promptly as the evaluation of estimate of fitness function.Wherein, the value of independent variable is chosen from termination set, unrestricted condition.
Existing technology has a lot of modes to select to carry out the object of genetic manipulation, and selection here adopts the mode of roulette to select to carry out the parent of genetic manipulation.Make
promptly the fitness function value of i individuals account for total fitness function value number percent; Accumulation probability
generates one [0; 1] the random number r in interval; The generation of r is to utilize random number generator to generate; Promptly specify this scope of 0 to 1, get a number at random in this scope the inside.The process of this peek confirmed by system, can not human intervention.If r is at interval [q
I-1, q
i] in, then the i individuals is selected.Be the parent individual replicate offspring individual directly.For example, for this colony shown in Figure 3, what at first will carry out is selection operation.Here select through roulette algorithm mentioned above.Here the fitness function value that can suppose this four stalk tree is respectively 0.4,0.3,0.2,0.1, and (the numeral here only plays signal, does not have physical meaning then at first to calculate fitness function value summation and be 1; The convenience in order to explain just), the fitness value that calculates every stalk tree then shared number percent in the fitness value summation is respectively 40%, 30%, 20%; 10%, the shared fitness value interval of then every stalk tree is [0,0.4], [0.4,0.7]; [0.7,0,9], [0.9,1].It is of future generation to select two stalks tree to get into, and then utilizes random number generator, generates random number twice; For the first time generate 0.3, in the fitness value interval of first stalk tree, then select first stalk tree; For the second time generate 0.5, in the fitness interval of second stalk tree, then select second stalk tree.To selecting first and second stalk tree execution replicate run, obtain progeny population as shown in Figure 4.
Then the progeny population that obtains through above-mentioned steps is carried out genetic manipulation, comprise and intersecting and mutation operation.For example, for two trees that provide among Fig. 4, interlace operation promptly is on the individual tree of 2 parents, to produce 2 point of crossing at random, and exchanging then with the point of crossing is the subtree of root node.By the experience of genetic algorithm, setting crossover probability here is 0.9.Here producing 2 point of crossing at random is node A and Node B, and the subtree that the probability exchange with 0.9 is a root node with these two nodes then can obtain the subtree of following Fig. 5.
Mutation operation promptly be select individuality in produce a change point at random; Then this point is set as root node initialization one stalk, setting the variation probability here is 0.05, and the variation probability is generally got within 0.1; Not too big, not so randomness is unfavorable for the search of solution space too by force.Probability with 0.05 carries out mutation operation to the node A of second stalk tree shown in Figure 5, and node A is by selected at random here.Subtree colony after the variation such as Fig. 6 are as showing.
Subtree colony after will passing through selection, intersect and making a variation is as new parent colony; Again carry out fitness evaluation and genetic manipulation; So developing repeatedly is equal to or greater than the operation that finished algorithm at 100 o'clock until the evolution iterations, and the population scale of here 100 and preceding text is irrelevant.Those of ordinary skill in the art should understand iterations and be not limited to 100, can number of iterations be set according to actual conditions and user's request in other embodiments.This moment, optimized individual was a net result.Can in solution space, search for optimum solution automatically through such genetic algorithm.At this moment optimized individual promptly is the result that prediction is come out, the wooden horse incident quantity that promptly will take place in the network.
Compared with prior art, the present invention has following two kinds of advantages:
Anticipation function to wooden horse incident generation quantity approaches better: based on the initial population of genetic algorithm from producing at random of binary tree modeling; Weigh individual quality with adaptive value; Adopt regeneration, intersection, mutation operation, the survival of the fittest of passing through some generations draws optimum solution or suboptimal solution.And what adopt is the layering of tree construction, the big or small scalable of tree.Such data representation mode and evolution mode are fit to wooden horse forecasting problem in the large scale network is found the solution, and find the potential rule of wooden horse incident generation quantity, therefore can well approach the function of wooden horse incident generation quantity.
Automatically suitable wooden horse incident generation quantitative forecast model is set up in search: because based on the thought of the genetic algorithms use auto-programming design of binary tree modeling; Computing machine can be searched for the proper model structure automatically; For the wooden horse forecasting problem in the large scale network; This method can be constructed rational forecast model automatically, and selects suitable parameters through heredity evolution constantly, reaches good prediction effect.
Though the present invention is described through preferred embodiment, yet the present invention is not limited to described embodiment here, also comprises various changes and the variation done without departing from the present invention.
Claims (7)
1. wooden horse event prediction method may further comprise the steps:
Step 1) is confirmed termination set and collection of functions, and said termination set comprises variable and constant, and the element in the said collection of functions then is to be used for operational symbol that the element in the termination set is operated;
Step 2) generate a series of function expressions at random according to termination set and collection of functions, said function expression is used to calculate current wooden horse incident generation quantity, and it is input as the quantity that the wooden horse incident takes place in the historical time section;
Step 3) is represented each function expression that the root node of said binary tree is in the collection of functions picked at random with the form of binary tree, intermediate node can be selected in collection of functions and termination set at random, and leaf node is selected in termination set at random;
Step 4) with whole binary tree colony as initial parent colony, to come execution genetic algorithm as evaluation to said initial parent colony to fitness according to the wooden horse quantity in the current network that said function expression was calculated and the difference of actual wooden horse quantity;
Step 5) utilizes the minimum function expression of the pairing said difference of the final optimized individual that generates of step 4) to come the generation quantity of the wooden horse incident in future in the network is predicted.
2. wooden horse event prediction method according to claim 1 wherein, also comprises the step that termination set and collection of functions are encoded before the step 3): each the element in termination set and the collection of functions is encoded with natural number.
3. wooden horse event prediction method according to claim 1, wherein, the depth capacity of every tree gets 4～6 in the step 3).
4. wooden horse event prediction method according to claim 1, wherein, step 4) may further comprise the steps:
Step 4-1) parent colony is carried out fitness evaluation, from parent colony, select to carry out the progeny population of genetic manipulation;
Step 4-2) to step 4-1) progeny population that obtains carries out interlace operation and mutation operation;
Step 4-3) with step 4-2) the binary tree colony that obtains is as parent colony, repeated execution of steps 4-1), 4-2), 4-3), till the multiplicity that satisfies appointment.
5. wooden horse event prediction method according to claim 4, wherein step 4-1) may further comprise the steps:
Adopt the method for roulette to come from parent colony, to select the high binary tree colony of fitness;
Every in choosing binary tree colony tree is carried out replicate run;
With the binary tree colony of being duplicated as progeny population.
6. crossover probability is 0.9 wooden horse event prediction method according to claim 4, wherein, step 4-2); The variation probability is 0.05.
7. multiplicity is 100 times wooden horse event prediction method according to claim 4, wherein, step 4-3).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101832196A CN102467635A (en) | 2011-07-01 | 2011-07-01 | Prediction method for trojan horse |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101832196A CN102467635A (en) | 2011-07-01 | 2011-07-01 | Prediction method for trojan horse |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102467635A true CN102467635A (en) | 2012-05-23 |
Family
ID=46071263
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011101832196A Pending CN102467635A (en) | 2011-07-01 | 2011-07-01 | Prediction method for trojan horse |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102467635A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101212681A (en) * | 2007-12-25 | 2008-07-02 | 海信集团有限公司 | Quick motion search method |
| CN101477224A (en) * | 2009-01-20 | 2009-07-08 | 南京航空航天大学 | Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning |
| CN101719193A (en) * | 2009-11-17 | 2010-06-02 | 上海电机学院 | Method for forecasting service life of brake |
-
2011
- 2011-07-01 CN CN2011101832196A patent/CN102467635A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101212681A (en) * | 2007-12-25 | 2008-07-02 | 海信集团有限公司 | Quick motion search method |
| CN101477224A (en) * | 2009-01-20 | 2009-07-08 | 南京航空航天大学 | Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning |
| CN101719193A (en) * | 2009-11-17 | 2010-06-02 | 上海电机学院 | Method for forecasting service life of brake |
Non-Patent Citations (1)
| Title |
|---|
| 张永强，陈华珊: "软件可靠性的GP演化建模", 《计算机工程》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kıran et al. | Swarm intelligence approaches to estimate electricity energy demand in Turkey | |
| Sahin et al. | Fault diagnosis for airplane engines using Bayesian networks and distributed particle swarm optimization | |
| Khalili-Damghani et al. | A hybrid fuzzy rule-based multi-criteria framework for sustainable project portfolio selection | |
| Chen et al. | Small-time scale network traffic prediction based on flexible neural tree | |
| CN107886161A (en) | A kind of global sensitivity analysis method for improving Complex Information System efficiency | |
| CN108764273A (en) | A kind of method, apparatus of data processing, terminal device and storage medium | |
| CN104298873A (en) | Attribute reduction method and mental state assessment method on the basis of genetic algorithm and rough set | |
| Mallipeddi et al. | Unit commitment–a survey and comparison of conventional and nature inspired algorithms | |
| CN109214503B (en) | Power transmission and transformation project cost prediction method based on KPCA-LA-RBM | |
| CN102075352A (en) | Method and device for predicting network user behavior | |
| Karimi-Majd et al. | A hybrid artificial immune network for detecting communities in complex networks | |
| Benjamin-Fink et al. | A road map for developing and applying object-oriented Bayesian networks to “WICKED” problems | |
| Dhamge et al. | Genetic algorithm driven ANN model for runoff estimation | |
| CN102456109A (en) | Training method for Trojan horse incident prediction least squares support vector machine, and prediction method | |
| Ishak et al. | Mining temporal reservoir data using sliding window technique | |
| Khan et al. | A multi-perspective revisit to the optimization methods of Neural Architecture Search and Hyper-parameter optimization for non-federated and federated learning environments | |
| CN117507365A (en) | Artificial intelligent control method based on dynamic self-adaptive parameter adjustment of materials and environment | |
| CN116883128A (en) | Method and device for excavating money laundering bulk, electronic equipment and computer storage medium | |
| Zahari et al. | Evaluation of sustainable development indicators with fuzzy TOPSIS based on subjective and objective weights | |
| CN109859062A (en) | A kind of community discovery analysis method of combination depth sparse coding device and quasi-Newton method | |
| CN115357568A (en) | Method, device, processor and storage medium for realizing social information discovery based on time-space domain credible graph neural network calculation | |
| CN102467635A (en) | Prediction method for trojan horse | |
| Huo et al. | An improved artificial bee colony algorithm for numerical functions | |
| Ashbolt et al. | A framework for short-term operational planning for water grids | |
| Jain et al. | Water scarcity prediction for global region using machine learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20120523 |