CN110750689A - Multi-graph fusion method - Google Patents
Multi-graph fusion method Download PDFInfo
- Publication number
- CN110750689A CN110750689A CN201911044229.4A CN201911044229A CN110750689A CN 110750689 A CN110750689 A CN 110750689A CN 201911044229 A CN201911044229 A CN 201911044229A CN 110750689 A CN110750689 A CN 110750689A
- Authority
- CN
- China
- Prior art keywords
- graph
- fusion
- fused
- graphs
- node
- 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
- 238000007500 overflow downdraw method Methods 0.000 title claims abstract description 33
- 230000004927 fusion Effects 0.000 claims abstract description 179
- 238000012545 processing Methods 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000013507 mapping Methods 0.000 claims description 15
- 238000007499 fusion processing Methods 0.000 claims description 10
- 230000004931 aggregating effect Effects 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 8
- 230000009191 jumping Effects 0.000 claims description 4
- 230000002194 synthesizing effect Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 17
- 230000002457 bidirectional effect Effects 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000006916 protein interaction Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a multi-graph fusion method, which comprises the steps of receiving a group of graphs as fused graphs, preprocessing the group of fused graphs, converting each fused graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, inputting subsequent processing activities, generating a group of multi-graph fusion schemes in a random mode to serve as an initial parent population, calculating the information entropy of the multi-graph fusion schemes in the initial parent population, forming the fitness of the multi-graph fusion schemes by the information entropy of the multi-graph fusion schemes, selecting a certain point on a shortest editing path between two multi-graph fusion schemes to be crossed as a crossing result of the two fusion schemes, realizing the simultaneous fusion of the multiple graphs and improving the fusion quality of the multiple graphs.
Description
Technical Field
The invention relates to the technical field of multi-image fusion, in particular to a multi-image fusion method.
Background
Graph (Graph) data is widely presented in various problem domains, such as protein interaction graphs in the biological domain, knowledge maps in the domain of knowledge representation, various structured products in the domain of software development, Graph-based databases in the domain of data storage. Any structured information can be represented in a graphical manner. A graph includes a set of nodes representing solid type information and a set of edges between the nodes representing relational information. The aim of graph fusion is to determine nodes with the same or similar semantics in different graphs, so as to eliminate redundant information in multi-graph data or transfer knowledge between different graphs. The essence of the multi-graph fusion problem is an optimization problem, which is to find a high-quality multi-graph fusion scheme from a large-scale multi-graph fusion scheme space.
However, the existing multi-graph fusion method does not really realize the simultaneous fusion of multiple graphs, but converts the multi-graph fusion problem into a fusion problem between a group of two graphs, so that the result of multi-graph fusion is related to the solving sequence of the group of two-graph fusion problems, and a high-quality multi-graph fusion scheme cannot be provided.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a multi-graph fusion method.
In order to solve the technical problems, the invention adopts the following technical scheme:
a multi-graph fusion method comprises the following steps:
s0, receiving a group of graphs as fused graphs;
s1, preprocessing a group of fused graphs, converting each graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, and using the fused graph as the input of subsequent processing activities;
s2, generating a group of multi-graph fusion schemes in a random mode to serve as initial values of parent population;
s3, calculating the information entropy of the multi-graph fusion scheme for the multi-graph fusion scheme in the parent population;
s4, converting the information entropy of each multi-graph fusion scheme in the parent population into the fitness of the multi-graph fusion scheme;
s5, checking whether a multi-graph fusion scheme with zero information entropy exists in the parent population, if so, returning to the multi-graph fusion scheme, and terminating the current multi-graph fusion process;
s6, checking whether the current multi-graph fusion process meets the termination condition, if so, returning to the multi-graph fusion scheme with the highest fitness in the current parent population, and terminating the current multi-graph fusion process;
s7, arranging all multi-graph fusion schemes in the parent population according to the sequence of the fitness from big to small, selecting a plurality of multi-graph fusion schemes with the highest fitness ranking, and putting the multi-graph fusion schemes into the child population;
s8, determining a group of multi-image fusion scheme pairs to be crossed from the parent population based on the fitness of the fusion scheme;
s9, for each multi-graph fusion scheme to be crossed, selecting a certain point on a shortest editing path between the two multi-graph fusion schemes as a cross result of the two fusion schemes, and then putting the cross result into a child population;
s10, when all the multi-image fusion scheme pairs to be crossed are crossed, the parent population is emptied, all the multi-image fusion schemes in the child population are added into the parent population, then the child population is emptied, and then the process jumps to S3.
Wherein a fused graph comprises a set of nodes and a set of edges existing between the nodes; one node in the fused graph may have a type; a node in the fused graph may have value information, such as a numerical value, a set of numerical values, an enumerated value, a set of enumerated values, a piece of text information, and the like; an edge in the fused graph may have a type; one edge in the fused graph can be a directional edge or a non-directional edge;
further, in the step S0, for each multi-graph fusion scheme to be intersected, selecting a certain point on a shortest editing path between the two multi-graph fusion schemes as an intersection result of the two fusion schemes, specifically including the following steps:
for two fusion graphs corresponding to the two multi-graph fusion schemes to be crossed, calculating the weight of each pair of nodes from the two fusion graphs for each type of nodes to form a one-to-one mapping relation with the maximum weight sum between the two fusion graph nodes, forming the arrangement of the type of nodes between the two fusion graphs for each fused graph based on the one-to-one mapping relation, and converting the arrangement into a non-intersected annular representation mode;
based on the arrangement of all types of nodes of all fused graphs between the two fused graphs, calculating the shortest editing path between the two fused graphs, and calculating the editing times required by the current cross operation;
randomly selecting a permutation from the permutations of all types of nodes of all fused graphs between two fused graphs, randomly selecting a ring from the permutation and randomly selecting two elements in the ring, and exchanging the two elements to form an editing operation;
and repeatedly carrying out editing operation for a specified number of times, and restoring a corresponding multi-graph fusion scheme from the generated arrangement as a cross result of the two graphs to be fused.
Further, in step S3, for the multi-graph fusion scheme in the parent population, calculating an information entropy of the multi-graph fusion scheme, where the information entropy of the multi-graph fusion scheme forms a fitness of the multi-graph fusion scheme, specifically including the following steps:
forming a fusion map based on the fusion scheme;
for the power set of each type of edge-out set of each node in the fusion graph, calculating a probability value of each element in the power set, calculating similarity between each pair of elements in the power set, calculating information entropy generated by the type of edge-out of the node based on the probability value of the power set elements and the similarity of the power set element pairs, and adding the information entropy generated by all types of edge-out of the node to form the information entropy generated by the edge-out of the node;
for the power set of the incoming edge set of each type of each node in the fusion graph, calculating a probability value of each element in the power set, calculating similarity between each pair of elements in the power set, calculating information entropy generated by the incoming edge of the type of the node based on the probability value of the elements in the power set and the similarity of the pairs of elements in the power set, and adding the information entropy generated by all types of the node to form the information entropy generated by the incoming edge of the node;
adding the information entropy generated by the outgoing edge and the incoming edge of each node in the fusion graph to form the information entropy of the node;
and synthesizing the information entropy of the fusion graph based on the information entropy of each node in the fusion graph.
Further, in step S2, generating a group of multi-graph fusion schemes in a random manner as an initial value of the parent population, specifically including the following steps:
for each type of node, generating a table with the row number of the fused graph and the column number of the sum of the number of the type of nodes in all the fused graphs, establishing a one-to-one mapping relationship between the fused graphs and the table rows, for each fused graph, randomly placing all the type of nodes contained in the fused graph into the cells in the table rows associated with the fused graph, ensuring that at most one node is placed in each cell in the table rows, aggregating the nodes in all the cells in each column of the table to form a set, aggregating the sets corresponding to all the columns of the table to form a set, then removing the empty set elements in the set to form a multi-graph fusion scheme of the type of nodes, and aggregating the multi-graph fusion schemes of all the type of nodes to form a multi-graph fusion scheme;
and repeating the steps until the number of the generated multi-graph fusion schemes is equal to the number required by the parent population.
Compared with the prior art, the invention has the beneficial technical effects that:
the multi-graph fusion method comprises the steps of receiving a group of graphs as fused graphs, preprocessing the group of fused graphs, converting each fused graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, using the fused graph as input of subsequent processing activities, generating a group of multi-graph fusion schemes in a random mode to serve as initial values of a parent population, calculating information entropies of the multi-graph fusion schemes for the multi-graph fusion schemes in the parent population, forming the fitness of the multi-graph fusion schemes, and selecting a certain point on a shortest editing path between the two multi-graph fusion schemes to be crossed as a crossing result of the two fusion schemes. Not only can fuse a plurality of graphs simultaneously, but also greatly improves the fusion quality of the graphs.
Drawings
The invention is further illustrated in the following description with reference to the drawings.
FIG. 1 is a schematic illustration of a multi-map fusion method of the present invention;
FIG. 2 is a diagram illustrating a step S1 of the multi-map fusion method according to the present invention;
FIG. 3 is a diagram illustrating a multi-map fusion method step S2 according to the present invention;
FIG. 4 is a schematic diagram of a fusion scheme for randomly generating each type of node in the multi-graph fusion method according to the present invention;
FIG. 5 is a diagram illustrating a step S3 of the multi-map fusion method according to the present invention;
FIG. 6 is a schematic representation of the conversion of a multi-map fusion protocol to a fused map in the multi-map fusion process of the present invention;
FIG. 7 is a schematic diagram illustrating the calculation of information entropy of any node in a fused graph in the multi-graph fusion method according to the present invention;
FIG. 8 is a diagram illustrating the entropy of information generated by calculating each type of outgoing edge of the node in the multi-graph fusion method according to the present invention;
FIG. 9 is a diagram illustrating step S9 in the multi-map fusion method according to the present invention.
Detailed Description
Referring to fig. 1, fig. 1 is a schematic diagram of a multi-map fusion method according to the present invention, the multi-map fusion method includes:
s0, receiving a group of graphs as fused graphs;
s1, preprocessing a group of fused graphs, converting each graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, and taking the fused graph as the input of subsequent processing activities;
s2, generating a group of multi-graph fusion schemes in a random mode to serve as an initial parent population;
s3, calculating the information entropy of the multi-graph fusion scheme for the multi-graph fusion scheme in the parent population;
s4, converting the information entropy of each multi-graph fusion scheme in the parent population into the fitness of the multi-graph fusion scheme;
s5, checking whether a multi-graph fusion scheme with zero information entropy exists in the parent population, if so, returning to the multi-graph fusion scheme, and terminating the current multi-graph fusion process;
s6, checking whether the current multi-graph fusion process meets a termination condition, if so, returning to the multi-graph fusion scheme with the highest fitness in the current parent population, and terminating the current multi-graph fusion process;
s7, arranging all multi-graph fusion schemes in the parent population according to the sequence of the fitness from big to small, selecting a plurality of multi-graph fusion schemes with the highest fitness ranking, and putting the multi-graph fusion schemes into the child population;
s8, determining a group of multi-image fusion scheme pairs to be crossed from the parent population based on the fitness of the fusion scheme;
s9, selecting a certain point on the shortest editing path between the multiple image fusion schemes to be crossed as a crossing result of the two fusion schemes, and then putting the crossing result into a child population;
and S10, when all the multi-image fusion scheme pairs to be crossed are crossed, emptying the parent population, adding all the multi-image fusion schemes in the child population into the parent population, emptying the child population, and then jumping to S3.
Referring to fig. 2, fig. 2 is a schematic diagram of step S1 of the multi-graph fusion method of the present invention, in which a group of fused graphs is preprocessed to convert each graph into a graph with nodes having types, edges having types, and edges having directions, and the graph is used as an input of a subsequent processing activity, and the method includes the following steps:
s1.1, a first step, namely, for all nodes without types in all fused graphs, assigning the same special type for the nodes without the types, and ensuring that the type is not the type of any node with the type in all the fused graphs;
s1.2, a first step of assigning a same special type to all the edges without types in all the fused graphs and ensuring that the type is not any type of the edges with types in all the fused graphs;
s1.3, a first third step, adding two edges with directions and opposite directions between two nodes at two ends of any edge without directions in all the fused graphs, assigning the types of the two edges as the types of the edges without directions, and then deleting the edges without directions.
Referring to fig. 3, fig. 3 is a schematic diagram of a multi-map fusion method in step S2 according to the present invention, in which a group of multi-map fusion schemes is generated in a random manner, and the method specifically includes:
s2.1, in the second step, the number of parent population is predetermined;
s2.2, in the second step, randomly generating a multi-graph fusion scheme, which specifically comprises the following steps: s2.2.1, step 1, randomly generating a fusion scheme of each type of node; s2.2.2, step 2, aggregating the fusion schemes of all types of nodes together to form a multi-graph fusion scheme;
s2.3, placing the randomly generated multi-graph fusion scheme into a parent population;
s2.4, a second step, checking whether the parent population contains the multi-image fusion schemes with the predetermined number, and if the number is less than the predetermined number, jumping to the second step.
Referring to fig. 4, fig. 4 is a schematic diagram of a fusion scheme for randomly generating each type of node in the second step 1 of S2.2.1 in the multi-graph fusion method of the present invention, which specifically includes:
s2.2.1.1, step 1A, determining the quantity ROW of the fused graph, determining the total number COL of the type of nodes in all the fused graphs, generating a table with COL columns and ROW ROWs, and randomly establishing a one-to-one mapping relation between all the fused graphs and all the table ROWs;
s2.2.1.2, a second step 1B, for each fused graph, randomly placing all the nodes of the type contained in the fused graph into the cells in the table row associated with the fused graph, and ensuring that at most one node is placed in each cell in the table row;
s2.2.1.3, a second 1C step, namely, aggregating the nodes in all the cells of each column of the table together to form a set, aggregating the sets corresponding to all the columns of the table together to form a set, then removing the empty set elements in the set, and taking the set obtained after removing the empty set elements as the fusion scheme of the type of nodes;
referring to fig. 5, fig. 5 is a schematic diagram of calculating information entropy of a multi-map fusion scheme for multi-map fusion schemes in a parent population in step S3 of the multi-map fusion method in the present invention, including:
s3.1, in the third step, converting a multi-graph fusion scheme into a fusion graph;
s3.2, thirdly, calculating the information entropy of any node in the fusion graph;
s3.3, a third step of calculating the information entropy for generating the fusion scheme according to the following formula
Referring to fig. 6, fig. 6 is a schematic diagram of a process of transforming a multi-map fusion scheme into a fusion map in the third step S3.1 of the multi-map fusion method of the present invention, which specifically includes:
s3.1.1, step III 1, taking each element contained in the fusion scheme as a node in the fusion graph;
s3.1.2, a third step, for any pair of nodes V0 and V1 in the fused graph, if V0 includes a node V0 from the fused graph X, and if V1 also includes a node V1 from the fused graph X, and there is an edge of type T pointing from V0 between V0 and V1, if there is no edge of type T pointing from V1 to V1 between V0 and V1, then an edge of type T pointing from V1 is added to the fused graph, and at the same time, a bidirectional association is established between the edge of type T pointing from V1 to V1 and the newly added edge of type T pointing from V1 to V1, so that given the information of any one of the two edges, information of the other edge can be obtained, but if there is an edge of type T pointing from V1 already between V1 and the T-type edge of type T, and there is a bidirectional association between the V1 and the V1-type T-type edge pointing from V1 and the V1 is a bidirectional association is established between the V1 and the V1-type T-type edge pointing from V1 and the newly added edge of type T-type is already exists between V1, and the two sides of type of the two sides pointing from V1 and the two points to the Association, a bidirectional association is established between the edge which points from V0 to V1 and is of the type T and the existing edge which points from V0 to V1 and is of the type T, so that the information of any one of the two edges can be given to obtain the information of the other edge;
s3.1.3, step III 3, repeating step III 2 until any step III 2 is executed without adding any new information to the fused graph.
Referring to fig. 7, fig. 7 is a schematic diagram of calculating the information entropy of any node in the fused graph in the step S3.2 and the step three in the multi-graph fusion method of the present invention, which specifically includes:
s3.2.1, step 1, calculating the information entropy generated by the edge of the node, including:
s3.2.1.1, step 1A, calculating the information entropy generated by each type of edge of the node.
S3.2.1.2, and step 1B, adding the information entropies generated by all types of outgoing edges of the node to obtain the information entropy generated by the outgoing edge of the node.
S3.2.2, step 2, calculating the information entropy generated by the entry edge of the node, which specifically includes: s3.2.2.1, step 2A, calculating the information entropy generated by each type of edge of the node; s3.2.2.2, step 2B, adding the information entropies generated by all types of the incoming edges of the node to obtain the information entropy generated by the incoming edge of the node.
S3.2.3, step three 3, adding the information entropy generated by the node edge-out and the information entropy generated by the node edge-in to obtain the information entropy of the node.
Referring to fig. 8, fig. 8 is a schematic diagram of calculating information entropy generated by each type of outgoing edge of the node in steps S3.2.1.1 and triethyl 1A in the multi-graph fusion method of the present invention, which specifically includes:
s3.2.1.1.1, a step of triethyl 1A, wherein for each outgoing edge of the node, a group of edges from the fused graph associated with the edge is obtained through the bidirectional dependency relationship established in the step of the third step 2, and further a group of fused graphs where the group of edges from the fused graph are located is obtained, which is called a fused graph set associated with one edge in the fused graph;
s3.2.1.1.2, a step III 1A, taking all the outgoing edges of the type of the node as a Set, and generating a Power Set (Power Set) of the Set, which is called as the Power Set of the outgoing edges of the specific type on the node in the fusion graph;
s3.2.1.1.3, and a third step 1A, for each element in the power set of the type of edge out of the node, obtaining all fused graphs which contain all the edges in the element but do not contain other edges in all the type of edge out sets of the node, and forming the fused graph set associated with the power set element;
s3.2.1.1.4, a step III 1A, for each element in the power set of the type of edge of the node, calculating a value generated by dividing the number of the element in the fused graph set associated with the element by the number of all fused graph nodes contained in the node, and referring the value as the probability value of the element in the power set;
s3.2.1.1.5, and a third step 1A, calculating a value generated by dividing the number of the intersection of the pair of elements by the number of the union of the pair of elements for each pair of elements in the power set of the type of the edge of the node, and calling the value as the similarity of the pair of elements in the power set;
s3.2.1.1.6, and step 1A, according to the following formula, calculating the information entropy generated by this type of edge of the node.
The implementation process of the step triethyl 2A of the multi-graph fusion method is similar to the implementation process of the step triethyl 1A, and is not described again.
Referring to fig. 9, fig. 9 is a schematic diagram of step S9 in the multi-graph fusion method according to the present invention, where, for each multi-graph fusion scheme to be intersected, a certain point on a shortest edit path between the two multi-graph fusion schemes is selected as an intersection result of the two fusion schemes, and then the intersection result is placed in a child population, and the schematic diagram specifically includes:
s9.1, recording the pair of multi-graph fusion schemes to be crossed as MS0 and MS 1;
s9.2, a ninth step, namely processing each type of node with the type of T, comprising the following steps:
s9.2.1, a ninth step B1, respectively obtaining sets of all elements with type T in multi-map fusion schemes MS0 and MS1, which are marked as Vset0 and Vset1, and ensuring that the number of the elements in the Vset0 is not more than the number of the elements in the Vset1, otherwise, the condition that the number of the elements in the Vset0 is not more than the number of the elements in the Vset1 can be met after the names of the two sets are exchanged;
s9.2.2, ninth step B2, if the number of elements in Vset0 is less than the number of elements in Vset1, adding a number of dummy fused graph nodes with globally unique identifiers to Vset0 (each dummy fused graph node is of type T, and each dummy fused graph node is now an empty set, not containing nodes from the fused graph), so that the number of elements in Vset0 equals the number of elements in Vset 1;
s9.2.3, ninth Eb 3, calculating the number of elements in the intersection of a pair of elements V0 and V1 from Vset0 and Vset1 as the weight of the pair of elements V0 and V1;
s9.2.4, finding a one-to-one mapping relation between all elements in the Vset0 and the Vset1 by adopting a Maximum assignment (Maximum assignment) method in a ninth step 4, so that the sum of the weights of all element pairs in the one-to-one mapping relation has the Maximum value in all one-to-one mapping relations between all elements in the Vset0 and the Vset1, and the one-to-one mapping relation is called as the Maximum mapping relation between the Vset0 and the Vset 1;
s9.2.5, ninth step B5, processing each fused graph G, including: s9.2.5.1, and a ninth step B5A, acquiring a set of all nodes with the type T in the fusion graph, and marking as vset; s9.2.5.2, a ninth step B5B, adding a set of dummy fused graph nodes of type T with globally unique identifiers to Vset so that the number of elements in Vset equals the number of elements in Vset0, randomly adding these dummy fused graph nodes to elements in Vset0 that do not contain a node from fused graph G, and ensuring that any one element in Vset0 contains and contains only one node from fused graph G, simultaneously randomly adding these dummy fused graph nodes to elements in Vset1 that do not contain a node from fused graph G, and ensuring that any one element in Vset1 contains and contains only one node from fused graph G; s9.2.5.3, a ninth step 5B, numbering all elements in vset by using continuous natural numbers after random sequencing, wherein the numbering range is 1, 2, 3, … and N, and N is equal to the number of the elements in vset; s9.2.5.4, a ninth step b 5D, based on the maximum mapping relationship between Vset0 and Vset1, the inclusion relationship of the fused graph node pair Vset0 and Vset1 for the nodes in Vset, and the number value of each element in Vset, obtaining a one-to-one mapping relationship between the set {1, 2, 3, …, N } and the set {1, 2, 3, …, N } elements, then converting this one-to-one mapping relationship into an arrangement with respect to natural numbers 1, 2, 3, …, N, referring to the arrangement of the T-type nodes in graph G on the multi-graph fusion scheme MS0 and MS1, and then representing this arrangement as a disjoint ring manner C1C2 … Ck, where k is the number of rings in this arrangement, and Ci (i =1, 2, …, k) represents one ring in this arrangement; s9.2.5.5, a ninth step 5E, for the arrangement C1C2 … Ck of T type nodes on the multi-graph fusion scheme MS0 and MS1 in the graph G, checking whether each ring Ci contains two or more numbers belonging to dummy fusion graph nodes in vset, if two or more numbers exist, randomly selecting two numbers from the ring Ci, exchanging the two numbers to obtain an updated arrangement represented in a disjoint ring manner, and continuing checking, exchanging and updating in the ninth step 5E in the updated arrangement until all rings in the updated arrangement do not contain two or more numbers belonging to dummy fusion graph nodes in vset;
s9.3, a ninth step of calculating the length of the shortest edit path between the multi-graph fusion schemes MS0 and MS1 by adding the numbers of elements included in the arrangement of all types of nodes in all fused graphs on the multi-graph fusion schemes MS0 and MS1, and then subtracting the sum of the numbers of loops included in the arrangement of all types of nodes in all fused graphs on the multi-graph fusion schemes MS0 and MS1 from this value produced by the addition;
s9.4, in the ninth step, uniformly adjusting the arrangement directions of all types of nodes in all fused graphs on the multi-graph fusion schemes MS0 and MS1 from MS0 to MS 1;
s9.5, ninthly, determining a positive integer of steps, where there are many values of steps, for example, a value between 0 and the shortest edit path length between the fusion schemes MS0 and MS1 may be randomly selected as the value of steps, a half of the shortest edit path length between the fusion schemes MS0 and MS1 may be selected as the value of steps, the information entropy of the fusion schemes MS0 and MS1 may be used as the weight, a specific proportion of the shortest edit path length between MS0 and MS1 may be selected as the value of steps, and it is ensured that if the information entropy of MS0 is greater than or equal to the information entropy of MS1, the value of steps is greater than or equal to the half of the shortest edit path length between MS0 and MS1, and it is ensured that if the information entropy of MS0 is less than or equal to the information entropy of MS1, the value of steps is less than or equal to the half of the shortest edit path length between MS0 and MS1, and so on the like.
S9.6, a ninth step, if the value of steps is zero, jumping to the next step, otherwise, randomly selecting one of all the arrangements generated in the ninth step, randomly selecting a ring in the arrangement, randomly selecting two elements in the ring, and then exchanging the two elements, thereby realizing the updating of the output result generated in the ninth step, and then repeatedly executing the steps-1 ninth step on the basis of the updating result;
s9.7, a ninth step of converting all permutations generated by performing steps for the ninth step into a corresponding multi-map fusion scheme, which is a multi-map fusion scheme of the shortest edit path between the multi-map fusion scheme MS0 and MS1, i.e., a crossover result of the multi-map fusion scheme MS0 and MS1, and then putting the crossover result into the offspring population.
The multi-graph fusion method comprises the steps of receiving a group of graphs as fused graphs, preprocessing the group of fused graphs, converting each fused graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, using the fused graph as input of subsequent processing activities, generating a group of multi-graph fusion schemes in a random mode to serve as an initial parent population, calculating the information entropy of the multi-graph fusion schemes in the parent population, forming the fitness of the multi-graph fusion schemes by using the information entropy of the multi-graph fusion schemes, and selecting a certain point on a shortest editing path between two multi-graph fusion schemes to be crossed as a crossing result of the two fusion schemes. The multi-image fusion method can simultaneously fuse a plurality of images, and improve the quality of multi-image fusion.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (4)
1. A multi-graph fusion method is characterized by comprising the following steps:
s0, receiving a group of graphs as fused graphs;
s1, preprocessing a group of fused graphs, converting each graph into a fused graph with a node having a type, an edge having a type and an edge having a direction, and using the fused graph as the input of subsequent processing activities;
s2, generating a group of multi-graph fusion schemes in a random mode to serve as initial values of parent population;
s3, calculating the information entropy of the multi-graph fusion scheme for the multi-graph fusion scheme in the parent population;
s4, converting the information entropy of each multi-graph fusion scheme in the parent population into the fitness of the multi-graph fusion scheme;
s5, checking whether a multi-graph fusion scheme with zero information entropy exists in the parent population, if so, returning to the multi-graph fusion scheme, and terminating the current multi-graph fusion process;
s6, checking whether the current multi-graph fusion process meets the termination condition, if so, returning to the multi-graph fusion scheme with the highest fitness in the current parent population, and terminating the current multi-graph fusion process;
s7, arranging all multi-graph fusion schemes in the parent population according to the sequence of the fitness from big to small, selecting a plurality of multi-graph fusion schemes with the highest fitness ranking, and putting the multi-graph fusion schemes into the child population;
s8, determining a group of multi-image fusion scheme pairs to be crossed from the parent population based on the fitness of the fusion scheme;
s9, for each multi-graph fusion scheme to be crossed, selecting a certain point on a shortest editing path between the two multi-graph fusion schemes as a cross result of the two fusion schemes, and then putting the cross result into a child population;
and S10, after all the multi-image fusion scheme pairs to be crossed are crossed, emptying the parent population, adding all the multi-image fusion schemes in the child population into the initial parent population, emptying the child population, and then jumping to S3.
2. The multi-graph fusion method according to claim 1, characterized in that: in step S9, for each multi-graph fusion scheme to be intersected, selecting a certain point on a shortest edit path between the two multi-graph fusion schemes as an intersection result of the two fusion schemes, specifically including the following steps:
for two fusion graphs corresponding to the two multi-graph fusion schemes to be crossed, calculating the weight of each pair of nodes from the two fusion graphs for each type of nodes to form a one-to-one mapping relation with the maximum weight sum between the two fusion graph nodes, forming the arrangement of the type of nodes between the two fusion graphs for each fused graph based on the one-to-one mapping relation, and converting the arrangement into a non-intersected annular representation mode;
based on the arrangement of all types of nodes of all fused graphs between the two fused graphs, calculating the shortest editing path between the two fused graphs, and calculating the editing times required by the current cross operation;
randomly selecting a permutation from the permutations of all types of nodes of all fused graphs between two fused graphs, randomly selecting a ring from the permutation and randomly selecting two elements in the ring, and exchanging the two elements to form an editing operation;
and repeatedly carrying out editing operation for a specified number of times, and restoring a corresponding multi-graph fusion scheme from the generated arrangement as a cross result of the two graphs to be fused.
3. The multi-graph fusion method according to claim 1, characterized in that: in step S3, for the multi-graph fusion scheme in the parent population, calculating the information entropy of the multi-graph fusion scheme, where the information entropy of the multi-graph fusion scheme forms the fitness of the multi-graph fusion scheme, specifically including the following steps:
forming a fusion map based on the fusion scheme;
for the power set of each type of edge-out set of each node in the fusion graph, calculating a probability value of each element in the power set, calculating similarity between each pair of elements in the power set, calculating information entropy generated by the type of edge-out of the node based on the probability value of the power set elements and the similarity of the power set element pairs, and adding the information entropy generated by all types of edge-out of the node to form the information entropy generated by the edge-out of the node;
for the power set of the incoming edge set of each type of each node in the fusion graph, calculating a probability value of each element in the power set, calculating similarity between each pair of elements in the power set, calculating information entropy generated by the incoming edge of the type of the node based on the probability value of the elements in the power set and the similarity of the pairs of elements in the power set, and adding the information entropy generated by all types of the node to form the information entropy generated by the incoming edge of the node;
adding the information entropy generated by the outgoing edge and the incoming edge of each node in the fusion graph to form the information entropy of the node;
and synthesizing the information entropy of the fusion graph based on the information entropy of each node in the fusion graph.
4. The multi-graph fusion method according to claim 1, characterized in that: in step S2, a group of multi-graph fusion schemes is generated in a random manner as an initial value of a parent population, and the method specifically includes the following steps:
for each type of node, generating a table with the row number of the fused graph and the column number of the sum of the number of the type of nodes in all the fused graphs, establishing a one-to-one mapping relationship between the fused graphs and the table rows, for each fused graph, randomly placing all the type of nodes contained in the fused graph into the cells in the table rows associated with the fused graph, ensuring that at most one node is placed in each cell in the table rows, aggregating the nodes in all the cells in each column of the table to form a set, aggregating the sets corresponding to all the columns of the table to form a set, then removing the empty set elements in the set to form a multi-graph fusion scheme of the type of nodes, and aggregating the multi-graph fusion schemes of all the type of nodes to form a multi-graph fusion scheme;
and repeating the steps until the number of the generated multi-graph fusion schemes is equal to the number required by the parent population.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911044229.4A CN110750689A (en) | 2019-10-30 | 2019-10-30 | Multi-graph fusion method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911044229.4A CN110750689A (en) | 2019-10-30 | 2019-10-30 | Multi-graph fusion method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110750689A true CN110750689A (en) | 2020-02-04 |
Family
ID=69281207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911044229.4A Pending CN110750689A (en) | 2019-10-30 | 2019-10-30 | Multi-graph fusion method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110750689A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102063715A (en) * | 2010-12-25 | 2011-05-18 | 浙江师范大学 | Method for fusing typhoon cloud pictures based on NSCT (Nonsubsampled Controurlet Transformation) and particle swarm optimization algorithm |
| CN102426606A (en) * | 2011-11-11 | 2012-04-25 | 南京财经大学 | Method for retrieving multi-feature image based on particle swarm algorithm |
| CN102508909A (en) * | 2011-11-11 | 2012-06-20 | 苏州大学 | Image retrieval method based on multiple intelligent algorithms and image fusion technology |
| CN103903055A (en) * | 2014-03-21 | 2014-07-02 | 国家电网公司 | Network reconstitution genetic algorithm based on all spanning trees of undirected graph |
| CN105279747A (en) * | 2015-11-25 | 2016-01-27 | 中北大学 | Infrared polarization and light intensity image fusing method guided by multi-feature objective function |
| JP2016032289A (en) * | 2014-07-25 | 2016-03-07 | 日本電気株式会社 | Image synthesis system, image synthesis method, image synthesis program |
| US20170006349A1 (en) * | 2015-06-30 | 2017-01-05 | Home Box Office, Inc. | Image providing service |
| CN106447612A (en) * | 2016-09-21 | 2017-02-22 | 湖南子午天地科技文化发展有限公司 | Image splicing method and image splicing generator |
| CN109408734A (en) * | 2018-09-28 | 2019-03-01 | 嘉兴学院 | A kind of collaborative filtering recommending method of fuse information Entropy conformability degree and dynamic trust |
-
2019
- 2019-10-30 CN CN201911044229.4A patent/CN110750689A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102063715A (en) * | 2010-12-25 | 2011-05-18 | 浙江师范大学 | Method for fusing typhoon cloud pictures based on NSCT (Nonsubsampled Controurlet Transformation) and particle swarm optimization algorithm |
| CN102426606A (en) * | 2011-11-11 | 2012-04-25 | 南京财经大学 | Method for retrieving multi-feature image based on particle swarm algorithm |
| CN102508909A (en) * | 2011-11-11 | 2012-06-20 | 苏州大学 | Image retrieval method based on multiple intelligent algorithms and image fusion technology |
| CN103903055A (en) * | 2014-03-21 | 2014-07-02 | 国家电网公司 | Network reconstitution genetic algorithm based on all spanning trees of undirected graph |
| JP2016032289A (en) * | 2014-07-25 | 2016-03-07 | 日本電気株式会社 | Image synthesis system, image synthesis method, image synthesis program |
| US20170006349A1 (en) * | 2015-06-30 | 2017-01-05 | Home Box Office, Inc. | Image providing service |
| CN105279747A (en) * | 2015-11-25 | 2016-01-27 | 中北大学 | Infrared polarization and light intensity image fusing method guided by multi-feature objective function |
| CN106447612A (en) * | 2016-09-21 | 2017-02-22 | 湖南子午天地科技文化发展有限公司 | Image splicing method and image splicing generator |
| CN109408734A (en) * | 2018-09-28 | 2019-03-01 | 嘉兴学院 | A kind of collaborative filtering recommending method of fuse information Entropy conformability degree and dynamic trust |
Non-Patent Citations (2)
| Title |
|---|
| 丁文静: "多传感器信息融合目标识别算法研究" * |
| 周延平: "互联网图像搜索中的多模态融合" * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Barenboim et al. | Distributed graph coloring: Fundamentals and recent developments | |
| Hosage et al. | Discrete space location-allocation solutions from genetic algorithms | |
| Hristakeva et al. | Solving the 0-1 knapsack problem with genetic algorithms | |
| Gaspers et al. | Feedback vertex sets in tournaments | |
| Buschbom et al. | Evolutionary history of vegetative reproduction in Porpidia sl (lichen-forming Ascomycota) | |
| Fischer et al. | An investigation of the recoverable robust assignment problem | |
| Forman et al. | Projections of the Aldous chain on binary trees: Intertwining and consistency | |
| US11803529B2 (en) | Unbalanced binary tree construction method based on calculation of binary boundary value | |
| CN108614932B (en) | Edge graph-based linear flow overlapping community discovery method, system and storage medium | |
| CN105893601A (en) | Data comparison method | |
| CN110750689A (en) | Multi-graph fusion method | |
| Chudnovsky et al. | Square-free graphs with no induced fork | |
| Agrawal et al. | Grouping Students for Maximizing Learning from Peers. | |
| CN116307328A (en) | Greedy solving method for travel business problem | |
| Palmer et al. | At most 3.55 n stable matchings | |
| CN110569415B (en) | Microblog user community discovery method based on network representation learning and neighbor resultant force | |
| Füredi et al. | Inequalities for the first‐fit chromatic number | |
| Behzadi et al. | Developing a genetic algorithm for solving shortest path problem | |
| Aizouk et al. | Modelling conflicting individual preference: Target sequences and graph realization | |
| Erlendsson | Effective enumeration of permutation classes and their juxtapositions | |
| Gen et al. | A comparative study of tree encodings on spanning tree problems | |
| Schneck | New parameters for beyond-planar graphs | |
| Dairyko | On exponential domination of graphs | |
| Cairncross | Proper 3-colorings of cycles and hypercubes | |
| Sadiquali et al. | A face change of monophonic dominating sets by non-adjacency in graphs |
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 | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20231229 |
|
| AD01 | Patent right deemed abandoned |