CN110276113A - A kind of network structure prediction technique - Google Patents
A kind of network structure prediction technique Download PDFInfo
- Publication number
- CN110276113A CN110276113A CN201910500053.2A CN201910500053A CN110276113A CN 110276113 A CN110276113 A CN 110276113A CN 201910500053 A CN201910500053 A CN 201910500053A CN 110276113 A CN110276113 A CN 110276113A
- Authority
- CN
- China
- Prior art keywords
- network
- straight
- structure prediction
- constraint
- gradient
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The present invention relates to network deconvolution and algorithm optimization technical field for this hair, realize a kind of method based on the slave protein-structural constraints of network deconvolution to protein structure prediction.Its step include: analog information data in a network from transmittance process；By the process from transmitting, the enclosed process that straight-forward network generates observation grid is constructed；To whole network plus constraint, prevents our recovery network over-fitting and preferably meet the characteristic of live network；The form with Lagrange constraint is converted by problem, so that problem more preferably solves；The method declined by gradient, continuous iteration finally obtain the straight-forward network information of recovery.The method reduce the influence of indirect relationship, help to distinguish the amino acid pair for having direct relation.
Description
Technical field
The present invention relates to the technology of network deconvolution and algorithm optimization, in particular to a kind of network structure prediction technique.
Background technique
In recent years, with the fast development of Network Science, Network Science is widely adopted in various environment in recent years, net
Network also becomes a kind of method for efficiently being used to indicate relationship between node.The weight on side indicates section between its interior joint and node
The intensity of correlation between point.
However due in network information from transmission effects, the side that we observe includes many indirect dependences.
Due to the transitivity of information, there is no direct relations between two nodes, but due between them there is indirect relation, I
May take for that there are direct relations when observation.Moreover, the direct relation between node may be overestimated
Because of the influence of indirect relation.In addition, with the continuous increase of network size, the shadow of the diffusion of information of indirect relation in a network
Sound can become increasingly severe.Traditional technology is only limited to analyze the structure of particular network, this presence is very big
Limitation, meanwhile, all methods do not consider the property that network itself has, this to the validity of method there is also
Certain influence.
Summary of the invention
In order to solve the above technical problems, the invention proposes a kind of network structure prediction techniques, comprising:
Observation grid obtains network data；
In analog network information from transmittance process, construct the closed solutions to go wrong；
Network is applied and is constrained；
The form with Lagrange constraint is converted by problem；
The method declined by gradient, continuous iteration；
Obtain the direct network information.
After this method, the invention has the following advantages that the present invention is in the prior art to the side of problem to be solved
The constraint based on network self property is added on the basis of method, while it is bright with method mathematically by problem to convert band glug
The problem of day constraint, finally the solution of straight-forward network is iterated using the method that gradient declines, meets us until finding
Desired straight-forward network.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the present invention.
Specific embodiment
Network structure prediction technique proposed by the present invention as shown in Figure 1, comprising: Step 1: observation grid, obtains network number
According to；Step 2: in analog network information from transmittance process, construct by straight-forward network generate observation grid enclosed process；
It is constrained Step 3: applying to network；Step 4: the form with Lagrange constraint will be converted into the problem of restoring straight-forward network；
Step 5: the method declined by gradient, continuous iteration；Step 6: obtaining straight-forward network information.
Step 1: detailed process is as follows for step 2: setting straight-forward network data as D, set the network data observed
It is all symmetrical matrix for F, D and F, wherein FijCorrelation between representing matrix interior joint i and node j includes directly related property
And indirect correlation, DijIt indicates the directly related property between node i and node j, was transmitted certainly by information in analog network
Journey can obtain:N is the order that item is unfolded.
Step 3: detailed process is as follows for step 4: in order to make straight-forward network D have sparsity, we add network D
Constrain ‖ D ‖0, it is as few as possible in order to make to restore non-zero element number in network D, but L0 norm is difficult to solve, so applying to network
Be constrained to L1, i.e. ‖ D ‖1, because L1 norm is the optimal convex approximation of L0 norm；
Then the form with Lagrange constraint is converted by problem, asked so that the straight-forward network D for meeting constraint passes through letter
What is ceased is minimum from error between the network and observation grid F that transmittance process generates:α is regularization parameter, | | | |1For
The 1- norm of matrix.
Step 5: detailed process is as follows for step 6: carrying out gradient solution to L (D), i.e., be divided into two parts to it and ask respectively
It leads, obtains result:, I
For unit matrix, the form that item unlimited in gradient adds up is converted into a closed solutions by Taylor's formula:Straight-forward network D is initialized first, by constantly using gradient
Update is iterated to straight-forward network D, finally obtains straight-forward network D required for us.
Protection content of the invention is not limited to above embodiments.Without departing from the spirit and scope of the invention, originally
Field technical staff it is conceivable that variation and advantage be all included in the present invention, and with appended claims be protect
Protect range.
Claims (4)
1. a kind of network structure prediction technique characterized by comprising
Observation grid obtains network data；
In analog network information from transmittance process, construct by straight-forward network generate observation grid enclosed process；
Network is applied and is constrained；
The form with Lagrange constraint will be converted into the problem of restoring straight-forward network；
The method declined by gradient, continuous iteration；
The network information being restored.
2. network structure prediction technique according to claim 1, it is characterised in that:
Straight-forward network data are set as D, set the network data observed as F, D and F are symmetrical matrixes, wherein FijIndicate square
Correlation between battle array interior joint i and node j includes directly related property and indirect correlation, DijIndicate node i and node j it
Between directly related property, by analog network information from transmittance process, can obtain:N is the order that item is unfolded.
3. network structure prediction technique according to claim 1, it is characterised in that:
In order to make straight-forward network D have sparsity, we add constraint ‖ D ‖ to network D0, in order to make to restore non-zero element in network D
Number is as few as possible, but L0 norm is difficult to solve, so being constrained to L1, i.e. ‖ D ‖ to what network applied1, because L1 norm is
The optimal convex approximation of L0 norm；
Then the form with Lagrange constraint is converted by problem, asked so that the straight-forward network D for meeting constraint passes through information
It is minimum from error between the network and observation grid F that transmittance process generates:α is regularization parameter, | | | |1For
The 1- norm of matrix.
4. network structure prediction technique according to claim 1, it is characterised in that:
Gradient solution is carried out to L (D), i.e., is divided into two parts to it and carries out derivation respectively, obtain result:, I is unit
The form that item unlimited in gradient adds up is converted to a closed solutions by Taylor's formula by matrix:Straight-forward network D is initialized first, by constantly using gradient
Update is iterated to straight-forward network D, finally obtains straight-forward network D required for us.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910500053.2A CN110276113A (en) | 2019-06-11 | 2019-06-11 | A kind of network structure prediction technique |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910500053.2A CN110276113A (en) | 2019-06-11 | 2019-06-11 | A kind of network structure prediction technique |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110276113A true CN110276113A (en) | 2019-09-24 |
Family
ID=67960625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910500053.2A Pending CN110276113A (en) | 2019-06-11 | 2019-06-11 | A kind of network structure prediction technique |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110276113A (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140277599A1 (en) * | 2013-03-13 | 2014-09-18 | Oracle International Corporation | Innovative Approach to Distributed Energy Resource Scheduling |
| CN105490858A (en) * | 2015-12-15 | 2016-04-13 | 北京理工大学 | Dynamic link predication method of network structure |
| CN105930686A (en) * | 2016-07-05 | 2016-09-07 | 四川大学 | Secondary protein structureprediction method based on deep neural network |
| CN106548210A (en) * | 2016-10-31 | 2017-03-29 | 腾讯科技（深圳）有限公司 | Machine learning model training method and device |
| CN107341571A (en) * | 2017-06-27 | 2017-11-10 | 华中科技大学 | A kind of social network user behavior prediction method based on quantization social effectiveness |
| KR20170141317A (en) * | 2016-06-14 | 2017-12-26 | 재단법인 전통천연물기반 유전자동의보감 사업단 | Method for predictiong drugs having opposite effects on disease genes in a directed netrowk and apparatus thereof |
| US20180114113A1 (en) * | 2016-10-20 | 2018-04-26 | Uber Technologies, Inc. | Intelligent regularization of neural network architectures |
| US20180293505A1 (en) * | 2017-04-06 | 2018-10-11 | Universite Paris Descartes | Method for clustering nodes of a textual network taking into account textual content, computer-readable storage device and system implementing said method |
| CN108681521A (en) * | 2018-04-18 | 2018-10-19 | 大连理工大学 | A kind of Fast Field algorithm of Real-time solution Model Predictive Control rule |
| CN109523021A (en) * | 2018-09-28 | 2019-03-26 | 浙江工业大学 | A kind of dynamic network Structure Prediction Methods based on long memory network in short-term |
| CN109584557A (en) * | 2018-12-14 | 2019-04-05 | 北京工业大学 | A kind of traffic flow forecasting method based on dynamic Decomposition mode and matrix fill-in |
| CN109817276A (en) * | 2019-01-29 | 2019-05-28 | 鲁东大学 | A kind of secondary protein structure prediction method based on deep neural network |
-
2019
- 2019-06-11 CN CN201910500053.2A patent/CN110276113A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140277599A1 (en) * | 2013-03-13 | 2014-09-18 | Oracle International Corporation | Innovative Approach to Distributed Energy Resource Scheduling |
| CN105490858A (en) * | 2015-12-15 | 2016-04-13 | 北京理工大学 | Dynamic link predication method of network structure |
| KR20170141317A (en) * | 2016-06-14 | 2017-12-26 | 재단법인 전통천연물기반 유전자동의보감 사업단 | Method for predictiong drugs having opposite effects on disease genes in a directed netrowk and apparatus thereof |
| CN105930686A (en) * | 2016-07-05 | 2016-09-07 | 四川大学 | Secondary protein structureprediction method based on deep neural network |
| US20180114113A1 (en) * | 2016-10-20 | 2018-04-26 | Uber Technologies, Inc. | Intelligent regularization of neural network architectures |
| CN106548210A (en) * | 2016-10-31 | 2017-03-29 | 腾讯科技（深圳）有限公司 | Machine learning model training method and device |
| US20180293505A1 (en) * | 2017-04-06 | 2018-10-11 | Universite Paris Descartes | Method for clustering nodes of a textual network taking into account textual content, computer-readable storage device and system implementing said method |
| CN107341571A (en) * | 2017-06-27 | 2017-11-10 | 华中科技大学 | A kind of social network user behavior prediction method based on quantization social effectiveness |
| CN108681521A (en) * | 2018-04-18 | 2018-10-19 | 大连理工大学 | A kind of Fast Field algorithm of Real-time solution Model Predictive Control rule |
| CN109523021A (en) * | 2018-09-28 | 2019-03-26 | 浙江工业大学 | A kind of dynamic network Structure Prediction Methods based on long memory network in short-term |
| CN109584557A (en) * | 2018-12-14 | 2019-04-05 | 北京工业大学 | A kind of traffic flow forecasting method based on dynamic Decomposition mode and matrix fill-in |
| CN109817276A (en) * | 2019-01-29 | 2019-05-28 | 鲁东大学 | A kind of secondary protein structure prediction method based on deep neural network |
Non-Patent Citations (2)
| Title |
|---|
| Y.GUERMEUR ETC.: ""Improved performance in protein secondary structure prediction by inhomogeneous score combination"", 《BIOINFORMATICS》 * |
| 李太勇: ""网络效应与竞争策略"", 《企业管理纵横》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Banchi et al. | Modelling non-markovian quantum processes with recurrent neural networks | |
| Diab et al. | On and in the Gross–Neveu and O (N) models | |
| He et al. | Robust matrix completion via maximum correntropy criterion and half-quadratic optimization | |
| WO2022011681A1 (en) | Method for fusing knowledge graph based on iterative completion | |
| Guo et al. | Promptfl: Let federated participants cooperatively learn prompts instead of models-federated learning in age of foundation model | |
| Lu et al. | Binarized aggregated network with quantization: Flexible deep learning deployment for CSI feedback in massive MIMO systems | |
| CN109801269B (en) | Tongue fur physique classification method based on competitive extrusion and excitation neural network | |
| Pourzanjani et al. | Improving the identifiability of neural networks for Bayesian inference | |
| Zhang et al. | Weights and structure determination of artificial neuronets | |
| Novotný et al. | Asymptotic properties of quantum Markov chains | |
| CN104732566A (en) | Hyperspectral image compressive sensing method based on nonseparable sparse prior | |
| Wilson et al. | Photometric redshift estimation with galaxy morphology using self-organizing maps | |
| Liu et al. | Nontrivial periodic solution of a stochastic non-autonomous SISV epidemic model | |
| Susca et al. | Top eigenpair statistics for weighted sparse graphs | |
| Hong et al. | Uniqueness of solutions to the 3D quintic Gross–Pitaevskii hierarchy | |
| CN110276113A (en) | A kind of network structure prediction technique | |
| Wang et al. | NTK-SAP: Improving neural network pruning by aligning training dynamics | |
| Wang et al. | An image fusion algorithm based on lifting wavelet transform | |
| Yip et al. | To Sample or Not to Sample: Retrieving Exoplanetary Spectra with Variational Inference and Normalizing Flows | |
| Bui et al. | Nonconvex regularization for network slimming: compressing cnns even more | |
| Chen et al. | Online dictionary learning over distributed models | |
| Ye et al. | Improving M-SBL for joint sparse recovery using a subspace penalty | |
| Zhang et al. | Cross-domain recommendation with multi-auxiliary domains via consistent and selective cluster-level knowledge transfer | |
| Tian et al. | An ADMM-Newton-CNN numerical approach to a TV model for identifying discontinuous diffusion coefficients in elliptic equations: convex case with gradient observations | |
| Knopf | Convergence and stability of locally ℝ N-invariant solutions of Ricci flow |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190924 |
|
| RJ01 | Rejection of invention patent application after publication |