• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    Apparatus 10 is provided in a method of generating a data-domain sampled network 206. In an embodiment, the method according to the present invention may include defining (166) a data domain, providing (172) a data-domain space or sub-domain, and analyzing (174) the data in the data domain 100. It leads. Analysis 174 selects the relevant dimension or variable (175), evaluates the cycles per dynamic range (176) or determines (176), selects the interpolation method (177), and calculates the number of sample points in each individual dimension. Selecting. Provision 180 of data-domain network 206 includes applying 184 multidimensional sampling theory to native data domain 100. The wait calculation 188 by the interpolation module 212 specifies a data-domain network in the native domain 100.
    公开号:KR20010079857A
    申请号:KR1020017003493
    申请日:1999-09-16
    公开日:2001-08-22
    发明作者:제이 엘 스미스
    申请人:추후제출；웨버 스테이트 유니버시티；
    IPC主号:
    专利说明:

    Data-Domain Sampled Networks {DATA-DOMAIN SAMPLED NETWORK}
    [2] U.S. Patent 5,796,922 (name of the invention: Trainable State-Sampled Network Controller, registered August 18, 1998, 922) describes a very useful analysis technique. In addition to the matrix algebra methodology, it relies on very useful properties of state-sampled domains. For example, depending on the separate and independent nature of the variable in the state domain, a simplified system of equations can be formulated and easily solved. However, if the data are highly combined, the prediction of independence and segregation between variables is very inaccurate.
    [3] U.S. Patent '922 also relies on modifications to state-space domains and subsequent analysis. In general, such transformation into state-space simplifies the analysis. However, in a number of situations that are actually encountered, information about the coupling between dimensions is lost by the necessary modifications.
    [4] Another problem in U.S. Patent '922 is the "knowledge" about the form of the equation. In control systems, classical control theory provides a plethora of terms in known forms to form various configurations of hardware or other control environments. In other classes of problems encountered in reality, the form of the equation is not necessarily known. Moreover, in many cases, even when the form of the equation is known or the equation itself is known correctly, it is virtually difficult to solve the governing equations due to the complexity of the computation that is difficult to process.
    [5] Therefore, there is a need for a method that can adjust, even retrieve and interpret the association between different variables (dimensions) in a data-domain without requiring the independence of the variables. There is also a need for a method that does not require modifications, especially modifications that lose information from the original data-domain. There is a need for a simplified, actually single step method of mapping output or surface in multidimensional data space directly from input data without having to make complex or impossible calculations and knowing the deductive form of the adjustment equation.
    [6] Therefore, there is a need for a system that simply and quickly correlates inputs with outputs associated with data in the original domain without requiring intermediate modifications. This is not possible with conventional methods, the complexity making the problem more difficult. Numerical methods are preferred in which the computerized algorithm for approximation can be made sufficiently accurate or sufficiently accurate. Therefore, data can be maintained within the original domain and can also be quickly, precisely, continuously and simply related to any correlation between related data parameters (eg input and output, independent and dependent variables). I need a way.
    [1] TECHNICAL FIELD The present invention relates to data analysis, and more particularly, to a novel data analysis system that maps the correlation of data while retaining the data in the original data domain rather than transforming the data into another domain for manipulation. And to a method.
    [14] 1 is a block circuit diagram of an apparatus according to the present invention suitable for operation within a computer system via a network;
    [15] 2 is a circuit diagram of a sensor system illustrating combined data generation for a data domain sample network;
    [16] 3 is a circuit diagram of a data domain,
    [17] 4 is a circuit diagram of the data domain for a surface representing one parameter set or value set in the data domain of FIG.
    [18] 5 shows a cross curve at a constant value of a variable or dimension in the data domain,
    [19] 6 is a block circuit diagram of a process of creating and using a data domain sample network;
    [20] 7 is a block circuit diagram of a data structure for implementing the invention in a computer memory;
    [21] 8 is a block circuit diagram of a process and data structure for implementing the apparatus and method of FIGS. 1-7, and
    [22] 9 is a representative circuit diagram illustrating interpolation of points on an associated surface of an associated data domain by the apparatus and method of the present invention.
    [7] In view of the foregoing, a first object of the present invention is to move between dimensions of a data-domain and to correlate inputs and outputs in a solution space without losing information from the data-domain through transformations. To provide a method and apparatus.
    [8] It is a second object of the present invention to provide an efficient method and apparatus for fusing multidimensional data, combined data sets or data strings without the need for independence or separation between variables (dimensions of data-domains), and without guesswork. have.
    [9] It is a third object of the present invention to provide a method and apparatus for maintaining information of interdependent variables in different dimensions of a data-domain.
    [10] A fourth object of the present invention is to simplify the processing of data by correlating the data to provide useful relationships (e.g. solutions, input / output relationships) in a single algorithmic operation without losing any dimension of continuity of the data-domain, Analysis and so on.
    [11] It is a fifth object of the present invention to practice the foregoing without requiring a priori knowledge of the equations or forms of equations relating variables to one another.
    [12] Apparatus and methods according to the invention, specifically and broadly described herein, are described in detail in accordance with the above-mentioned objects. In an embodiment of the method and apparatus according to the present invention, a data network of data points within the original data-domain (as opposed to a hardware computer network in which it operates), without modification to a domain that loses important features of the data. The data can be manipulated and used. For example, binding, persistence of variables, and persistence of variable differentials are maintained.
    [13] Other objects and features of the present invention and the foregoing are described with reference to the accompanying drawings and become more apparent in the detailed description of the invention and the appended claims. The drawings merely implement exemplary embodiments of the invention and do not limit its scope.
    [23] It is to be understood that the components of the present invention may be arranged and designed in a wide variety of configurations, as generally described and illustrated in the drawings. 1 to 9, the detailed description of the embodiments of the systems and methods of the present invention is not intended to limit the scope of the present invention, which is as broad as claimed herein. The examples illustrate certain, preferred embodiments of the invention, which are best understood by reference to the drawings in which like parts are represented by like numerals.
    [24] It will be apparent to one skilled in the art that various modifications can be facilitated without departing from the spirit of the invention. Therefore, the detailed description of the drawings is only used as an example and illustrates preferred embodiments consistent with the claimed invention.
    [25] In FIG. 1, apparatus 10 provides one or more nodes 11 (client 11, computer 11) comprising a processor 12 or a CPU 12 in the present invention. All components exist as a single node 11 or as a plurality of nodes 11, 52 located apart from one another. The CPU 12 is easily connected to the memory device 14. Memory device 14 includes one or more devices, such as a hard drive, non-volatile storage 16, read-only memory (ROM) and random access (usually volatile) storage 20 (RAM).
    [26] The device 10 includes an input device 22 that receives input from a user or other device. Similarly, output device 24 may be provided within node 11 or accessible within device 10. A network card 26 (interface card) or port 28 is provided for connecting an external device such as the network 30.
    [27] Internally, the bus 32 is easily interconnected with the processor 12, the memory device 14, the input device 22, the output device 24, the network card 26 and the port 28. The bus 32 is considered a data carrier. As such, the bus 32 is realized in a number of configurations. Wired communication by visible light, infrared ray and radio frequency, optical fiber line communication, and wireless electromagnetic communication are likewise appropriately provided to the bus 32 and the network 30.
    [28] Input device 22 includes one or more physical embodiments. For example, the keyboard 34 is used for interaction with the user, such as the mouse 36 or the stylus pad 37. The touch screen 38, telephone 39 or telephone line 39 is used to communicate with other devices, users and the like. Similarly, scanner 40 is used to receive graphical inputs that are translated or not translated into other text formats. All types of memory devices 41 (hard drives, floppies, etc.) can be used as input devices, whether within or from other nodes 52 or nodes 11 of network 30 or from other networks 50. .
    [29] Similarly, output device 24 also includes one or more physical hardware units. For example, generally, port 28 receives input from node 11 and sends output. The monitor 42 provides output to the user for feedback during processing or to assist two-way communication between the processor 12 and the user. The printer 44 or the hard drive 46 is used to output information as the output device 24.
    [30] In general, the network 30 to which the node 11 is connected is also connected to other networks via a router 48. In general, two nodes 11, 52 are located on the network adjacent to the networks 30, 50, or as separate nodes on the internetwork 11, 52 by multiple routers and multiple networks 50. Are separated. Individual nodes 52 (eg, 11, 52, 54) have various communication possibilities.
    [31] In one embodiment, minimal logic is available to all nodes 52. The individual nodes 11, 52, 54 are referred to as node 11 or node 52, as when referring to all nodes together. Each node includes a processor 12 with some other components 14-44.
    [32] Network 30 includes one or more servers 54. The server is used to manage, store, communicate, transfer, access, update, etc. the actual number of files, databases, etc. to other nodes 52 on the network 30. The server is accessed by all nodes 11, 52 on the network 30. In general, in the present invention, nodes 11 and 52 that are accessible to information or files are referred to as servers. Thus, the "web site" available to the user of the internetwork 50 is considered a server. Other special functions, including communications, applications, directory services, etc., may be implemented by individual server 54 or multiple servers 54. Nodes 11 and 52 are servers 54.
    [33] Node 11 needs to communicate with server 54, router 48 or other nodes 52 via network 30. Similarly, node 11 needs to communicate via another network (such as or different from network 30) of Internetwork 50 connecting nodes 52 away from network 30. Similarly, individual components 12-46 need to communicate data with each other. In general, a communication link exists between two devices.
    [34] In FIG. 2, there is a system 60 for observing an object moving in the azimuth direction 64 and the up and down direction 66. As shown in FIG. 2, radiation 68 (image) proceeds from object 62, which is referred to as sun 62. In the illustrated embodiment, the sensor suite 70 includes detectors 72 and 74: azimuth sensor 72 and up and down sensors 74 that detect radiation (images) reflecting motion in a dimension perpendicular to each other. ).
    [35] Sensors 72 and 74 have connections 76 and 78 or connecting data lines 76 and 78 that connect sensors 72 and 74 to data acquisition system 80, respectively. The data acquisition system 80 performs digital signal processing or other preliminary processing. Alternatively, the data acquisition system 80 records the parameter output by each sensor 72, 74.
    [36] The data acquisition system 80 is connected to the external computer 11 by a cable 82 or other connection 82. The connection 82 outputs data from the data acquisition system 80 to the computer 11 and outputs control data from the computer 11 to the data acquisition system 80.
    [37] In general, the computer 11 is connected to the network 84 to provide raw, preprocessed, or fully analyzed data from the data acquisition system 80 to other nodes on the network 84. As a practical matter, as the network grows, the network 84 is a local area network or an internetwork to provide an input signal to the computer 11 controlling the data acquisition system 80 or simply to the computer 11. It is also a user of data provided by and displaying or reflecting data from the data acquisition system 80.
    [38] Each sensor 72, 74 has a "line of sight" of sights 86, 88 toward object 62. In the illustrated embodiment, the aperture system 90 includes a mask 92 and an orientation aperture 94 to protect data reflecting motion of the object in azimuth 64 and up-and-down 66. And an upper and lower apertures 96. Nevertheless, as a practical matter, the motion of the object in all directions 64, 66 passes through respective apertures 94, 96 and eventually radiation received by respective sensors 72, 74. Affects) As a result, the data recorded by the data acquisition system 80 for each sensor 72, 74 is virtually combined. In fact, all movement of the object 62 acts on the radiation 68 detected by both sensors 72, 74. The example of FIG. 2 is a simplified example shown in two dimensions. Other levels of system may exist.
    [39] In the method and apparatus according to the invention, the data provided by the sensors 72, 74 to the data acquisition system 80 need not lose the information stored in the association relationship. Mathematically, partial differential equations exist to explain phenomena where the variables or dimensions of the relevant space are not independent.
    [40] To the extent that data received on multiple channels of the data acquisition system is independent, information is not lost by lack of coupling between the channels or by estimating independence. However, in FIG. 2, if it is assumed that the data recorded by the two sensors 72, 74 in the data acquisition system 80 are separable by the channel, this is an inaccurate estimate. Storing and analyzing data and estimating independence by individual channels or separate modifications destroys the combined information.
    [41] Thus, no estimation of linearity or independence is necessary for the data. Instead, the data is maintained in the native domain 100 (Figure 3). By keeping data in the native domain 100, distortion or specificity, disconnection, etc. do not need to be introduced by modification. Instead, the data can be combined and recorded as detected, and the device 10 according to the present invention processes the data to determine correlations between all variables or dimensions of the data domain 100.
    [42] When referring to the data domain 100, people naturally think of independent and non-independent variables. In many cases, however, the independence and independence of variables is not known or even recognized. Thus, the apparatus and method according to the present invention has the advantage of preserving detectable information in the combination between the channels of the data acquisition system 80.
    [43] In FIG. 3, the data domain 100 may include a first variable 102 or a first dimension 102, a second variable 104 or a second dimension 104 and a third variable 106 or a third dimension 106. It is limited to the term. Since three or more dimensions are difficult or impossible to explain, FIG. 3 shows three dimensions. Nevertheless, there is no limit on the number of dimensions in the data domain 100.
    [44] Data domain 100 includes several points 108 of a surface defined by first dimension 102 and second dimension 104. Corresponding to each point 108 is the value 110 of the dimension. People naturally want to regard the first dimension 102 and the second dimension 104 as independent dimensions and the dimension 106 of the value 110 as a third non-independent dimension.
    [45] Nevertheless, in the present invention, all dimensions 102, 104, 106 are selected as functional dimensions or value dimensions and are sometimes referred to as solution dimensions or related functions 110. There is no need to estimate what is non-independent and what is independent in the stored data. Anything that can be detected and recorded is stored in the data domain 100. Actual dimension numbers 102, 104 and 106 are used. Thus, the number of actual variables 102, 104, 106 is used in the data domain 100.
    [46] In addition, the increments 112 and 114 are regular or irregular and may be known or unknown in advance. For example, when the data acquisition system 80 records data, the data is usually a column of time. Thus, for every channel of data acquisition system 80 some series of points 108, 110 are recorded with one value for each channel at a time common to all channels of the data sequence. Later, in processing, various sub-domains 116, 118 are defined for investigation or analysis, and the like. These sub-domains 116, 118 are defined by increments 112, 114 within each dimension 102, 104.
    [47] In FIG. 4, the data domain 100 of FIG. 3 appears to include a surface 120 connecting the values 110 in the dimension 107 or the functional dimension 106. Note that functional dimension 106 is an arbitrary name. As a practical matter, the functional dimension 106 is the proposed dimension by controlling the other variables 102 and 104. However, there is no need to have preconceived notions about the independence and independence of variables 102, 104 and 106, except as necessary to clarify by way of example.
    [48] Surface 120 extends in all dimensions 102, 104, 106. Surface dimension 122 is not the same as dimension 102. Likewise, surface dimension 124 is not the same as dimension 104. Rather, dimensions 122 and 124 are dimensions located along surface 120, so that surface 120 is in the data domain 100 from a direction 102, 104, that is, a surface defined by dimensions 102, 104. Projected.
    [49] In analyzing the surface 120 of the data domain 100, people pay attention to the local maximum value 126 and the local minimum value 128. By all means, there is an inflection point between all the maximum value 126 and the minimum value 128. In measuring the required precision, in order to accurately represent the data domain 100 by multi-dimensional sampling theory (hereinafter referred to as sampling theory), the calculation of the number of inflection regions 130 is necessary to measure the required number and degree of sampling points or interpolation function. Measure the parameters of the other functions needed for.
    [50] In FIG. 5, the first dimension 134 and the second dimension 136 correlate with and interrelate with the third dimension 138. In FIG. 5, the intersection curve 140 between the surfaces 142 represents the value of the third dimension 138 at any value of the first and second dimensions 134, 136.
    [51] The plain 144 of the dimension 136 corresponds to the fixed value 145 of the second dimension 136. Intersection 140 of plane 144 and surface 142 represents curve 140 in data domain 100 as a constant 145 of variable 136 or dimension 136.
    [52] The presence or absence of plane 144, that is, the presence of plane 144, does not necessarily mean that plane 144 is limited. That is, all variables 134, 136, 138 are interdependent. Changes in variables 134, 136, and 138 change the structure of surface 142. Among the many problems encountered in real life related to physical system data, it is impossible to define a surface or curve 140 without resorting to a numerical approximation system.
    [53] It requires a lot of computing power, and there are very complex relationships. The apparatus and method according to the invention define the relationship between a surface and a variable without resorting to modifications, estimations, independence or combinations, and without the need for highly complex and time-consuming calculations.
    [54] Distance 146 is considered a value 146 in dimension 136. Plane 144 is considered to be a series of points, which unfortunately exist on both the plane and the surface. The value 150 is a variable 136 that corresponds to any value 146 of the variable 136 that corresponds to the value of the zero variable 134 or the dimension 134 and the value of the dimension 138. Is represented by
    [55] Lattice 154 defines the sub-domains of data domain 100. The grating 154 is increased arbitrarily or evenly. In the present invention, the regularity of a variable or dimension is usually controllable but the time cannot actually be controlled. Rather, the variable time is increased, and the data acquisition system 80 is controlled to record the channel with a certain increase in time. Nevertheless, time is not really controlled. Thus, the data domain actually contains only one or none of the regularly increased dimensions, such as time. Other dimensions vary within the range of values of the parameters measured in these dimensions.
    [56] Nevertheless, in other systems, several parameters are provided or controlled as inputs in the data domain. The method and apparatus according to the present invention changes very rapidly from conventional state-sampled control networks, which usually perform state-domain increments in a regular manner. In the present invention, control of data is not necessary.
    [57] In FIG. 6, a process 160 or method 160 for creating 162 and using 164 a data-domain sampled network is shown. Initially, creation 162 of the data domain sampled network includes providing 165 and optionally defining 166 the data domain. Data domain 100 is defined by dimensions 134, 136, and 138, but need not be limited to the number of variables or dimensions on which data can be recorded. This includes the characteristics as well as the value of each unit or the value of each dimension. Defining step 168 includes selecting steps 167 and 169. Other processing steps are also deleted in the selected embodiment, but the bracket on the label indicates which process is optional.
    [58] Restriction step 167 is responsible for defining the independent variable domains. The defining step 169 is responsible for defining the scope of the function. The concept of domains and ranges, as well as independent variables (inputs) and non-independent variables (outputs or functions), ranges from arbitrary to absolute and complete. Nevertheless, the definition of the functional range includes parameters or dimensions 138 or related crystals, and the relevant surface 142 is preferably observed but not necessarily controlled.
    [59] Conversely, the definition 167 of the independent variable domain involves the selection of another dimension 134, 136, the effect of which is the value 142 or points () of the surface located within the surface 142 on the dimension 134, 136. It is desirable to parameterize, quantify, or quality in determining how to relate to, or affect, 148, 150).
    [60] Thus, delimitation step 168 is considered an arbitrary step because the relationship in the data domain is unique. This relationship is maintained and not destroyed by manipulation, deformation or the like as in the prior art.
    [61] Selecting a dimension or variant 170 for analysis is responsible for determining which data domain 100's dimension to depend on. Providing a point in the data-domain space 100 or the sub-space 100 is whether it is a non-independent, independent or unknown relationship, or provides related data.
    [62] For calculation, analyzing 174 the data in the data-domain spectrum includes selection 175 of individual dimensions. Thus, steps 175, 176, 177, 138 are repeated for each dimension of data-domain space 100 or data domain 100. Determining 176 the cycle for the dynamic range includes calculating the surfaces 120, 142 for the inflection 130. In general, the frequency, rate of change, maximum value 126 and minimum value 128 will affect not only the complexity of the interpolation system but also at least the number of data points needed for the sample size.
    [63] The choice of interpolation method 177 is arbitrary. Sampling theory and interpolation theory have developed the best technology. Selection of the interpolation method 177 is preferred to obtain access to the optimal interpolation method and function. This helps to determine the cycles per dynamic range in the relevant dimensions (134, 136, 138).
    [64] Selecting 178 the number of sample points in the relevant dimension 134 (the related dimension 134 will be used to represent the dimensions 134 and 136 in the space 100) is a direct function of the decision step 176. The process 179 initially begins with a selection step for all relevant dimensions 134 including all dimensions 106 and 138 of the function surfaces 120 and 142.
    [65] Providing a data-domain network, more appropriately a data-domain sampled network, begins by selecting an interpolation function. Applying 184 the sampling theory to native data domain 100 or data domain 100 suggests an interpolation function type as well as a specific type of optimal interpolation function. For example, dividing by sine ((x) / x) is called a sine function and provides a form of interpolation function that is appropriate and optimal. The interpolation function is similar to the interpolation function 186 of FIG. 6. In FIG. 6, interpolation function 186 associates a function with a series of wait function sums divided by the value of a data point or a function of a particular dimension 134 in data-domain 100.
    [66] On the other hand, the correlation between the functions in the various dimensions 134, 136, 138 or the values between the values in the various dimensions 134 (generally the dimension 134) is an important factor but is not estimated in advance, so the functionality is somewhat arbitrary. . Details of how to use the interpolation function are in the prior art, but are not bound to a particular solution in the present invention. Nevertheless, interpolation function 186 is known to be appropriate.
    [67] The calculation 188 (see FIG. 7) of the weight 208, denoted by "W" in the interpolation function, is done in a manner known in the prior art. Nevertheless, the above-mentioned prior art includes a method suitable for weight calculation 188.
    [68] The use of the data-domain sampled network 218 (see FIG. 9) includes the selection 190 of the relevant data-domain dimension 134. Regardless of which variable 134 or dimension 134 is independent from other dimensions 134 (eg, 136, 138, etc.). Selection 190 involves determining whether a parameter and its effects are desired to be observed. In general, people select a character that has a functional relationship with a parameter, and speak of a change in character merit as one of the values 106 in the data-domain 100.
    [69] Determination 192 of a random set of inputs and outputs involves selection 190 of the associated dimension 134. In general, inputs and outputs have meaning in experimental design. Nevertheless, in the apparatus or method according to the invention, the actions and relationships need not be controlled, modified or manipulated as a requirement for finding a solution. Thus, the set of arbitrary dimensions 134 included in the selection in the data domain 100 is selected. Nevertheless, to get the maximum information, all dimensions 134 (recall 134 represent some dimension and all dimensions) are used, and a particular dimension 134 is selected for observation as an output set. Similarly, specific dimensions 134 are located in the input set and are plotted or calculated in regular increments to accurately observe the output set.
    [70] Thereafter, selecting 194 input points in the data domain 100 is an iterative process. Selecting (194) any input point in the data domain 100 provides a point in the data-domain 100, wherein the value 106 of the interpolation function 186 in the data-domain is a "solution" corresponding to the relevant point. (solution) ".
    [71] Interpolating the output point 196 means calculating the interpolation value 110 corresponding to the point 108 in the selected dimensions 102 and 104 of the data domain 100. If any equation is available, the function value is found outside of the data domain. However, knowledge of the functional relationship between the data domain 100 and other parameters is not necessary. In general, related inputs and outputs are considered part of the data domain. Thus, typically, all values found during interpolation step 196 are within the bounds of the data domain. Similarly, interpolation step 196 finds dimension value 134 in data domain 100.
    [72] 7 shows operation data that can be executed by the present invention. Generally, computer readable memory device 196 corresponding to computer 11 stores various data structures 200-204. As a practical matter, as in the description of FIG. 6, the multidimensional sampling module 198 is responsible for implementing the sampling theory to provide an analysis of the data in the data-domain 100 spectrum.
    [73] Thus, the multidimensional data-domain analysis output 202, including frequency, cycles, interpolation method selection, number of samples required in each dimension, and the like, is determined by the sampling theory and output of the multidimensional sampling module 198. Is provided.
    [74] To provide the weight 208 or the weighting 188, the correlation module 204 uses the output from the multidimensional data-domain analysis 202 as well as the raw data point 200 stored in the memory 196. .
    [75] As described, the data-domain sampled network 206 and particularly the weight 208 reflect the information obtained at the origin point 200 of the data domain 100. In order to provide relevant values in the data domain 100, the interpolation module 212 depends on the weight 208 and the interpolation function 21 of the data-domain sampled network and the data point 200 of the data domain 100. have.
    [76] As described, the selection of input to output can be selected from the contents of the data domain 100. Thus, the relevant input or output values are derived from the data domain, but the actual raw material 200 is typically provided because the interpolation function 210 provides for all intermediate points located on the surfaces 120 and 142 of the data domain 100. Not in the value of).
    [77] 7, the data structure 194 of the memory device 196 is executed in a processor 215 such as the processor 12 of the computer 11. To provide multidimensional data-domain analysis 202, output 202, raw data point 200 is provided (216a) or processed (216a) by multidimensional sampling module 198 (216 is general, 216a-). 216h is specific). Thus, the number of cycles, the interpolation method and the number of samples required for optimization are all provided as part of the multidimensional data-domain analysis 202 or output 202.
    [78] An interrelated module 204 is provided to utilize the appropriate mechanism for correlating data points 200. An important advance with the present invention over the prior art is the fact that the interrelated module only depends on the data 200. Multidimensional data-domain analysis 202 is provided based solely on data domain 100 and data 200, not on transformations, a priori information, estimated or analyzed equations, separate channels, and the like. Thus, through the processing 216c, 216d and analysis output 202 of the correlating module 204 in the data 200, the data domain was exclusively transformed into another domain.
    [79] The data-domain sampled network 206 occurs as a direct result 216e of the correlation module 204. The resulting weight determines the characteristics of the data-domain sampled network 206. Thus, the weight 208 along with the data 200 is provided to the interpolation module 212 along with the interpolation function 210 provided 216h to the interpolation module 212. The interpolation module 212 depends on the data 200 and the weight 208 provided and performs interpolation by the provided interpolation function 210. The output of interpolation module 212 is an important value within data domain 100 for all points 226 (see FIG. 9) anywhere in data domain 100.
    [80] 9, data domain 100 is seen as a complete, persistent space. Due to interpolation module 212, point 110 or value 110 is continuously defined anywhere within domain 100 while providing surface 220. Surface 220 consists of a point 221 (eg, value 110) defined by vector 224. Generally, people consider the sub-domain 222 to be an independent space 222, while the surface 220 is considered to be a solution 220 or a non-independent surface 220.
    [81] Thus, the vector 224 corresponds to the point 221 or value 221 mapped through the sub-domain 222 of the data-domain 100. That is, data-domain 100 includes surface 220 and sub-domain 222. On the other hand, sub-domain 222 and surface 220 are considered separate, arbitrary inputs and outputs (eg, independent, non-independent) selection. Surface 220 is an indication of relevant parameters of data domain 100 where observation is desired.
    [82] Due to the interpolation function 210 and the weight 208, the points 226 that are not included in the original data point 223 but are present in the data domain 100 are provided as outputs by the interpolation module 212 to the vector 230. , Ie, provided as an output by the interpolation module 212, has a value 232 which is a point 232 on the surface 220.
    [83] As described above, the present invention relates to a novel data analysis system and method for mapping data correlations while retaining data in the original data domain rather than transforming the data into another domain for manipulation, an important feature of the data. The data can be manipulated and used in the data network of data points (as opposed to the hardware computer network on which it operates) within the original data-domain, without modification to the domain that loses the data. For example, binding, persistence of variables, and persistence of variable differentials are maintained.
    权利要求:
    Claims (10)
    [1" claim-type="Currently amended] For a data-domain sampled network,
    Provide a data domain,
    Provide a data point to the data domain,
    Multi-dimensional sampling theory analyzes the behavior of data points within the data domain to determine the sampling distribution to be applied to the data domain.
    Providing a data-domain sampled network based on the analysis output in the data domain.
    [2" claim-type="Currently amended] The method according to claim 1,
    Providing a computer readable memory device comprising an executable operation data structure,
    The data structure provides a multidimensional sampling module for analyzing a data point in a data domain to determine a sampling architecture.
    Processing a data point and sampling architecture to provide a data-domain sampled network that correlates the data points and provides a data-domain sampled network that reflects the interrelationship of the data. Way.
    [3" claim-type="Currently amended] The method according to claim 2,
    And providing an interpolation module for interpolating the values of the relevant first dimension of the data domain from the relevant points defined by parameters of the other dimensions of the data domain.
    [4" claim-type="Currently amended] The method according to claim 1,
    A data-domain sampled network method, further comprising providing an interpolation function deployed using data points in the data domain.
    [5" claim-type="Currently amended] The method according to claim 4,
    And providing an interpolation module for interpolating the values in the first dimension corresponding to the values in the dimension in which the second and third dimensions are combined.
    [6" claim-type="Currently amended] The method according to claim 5,
    Interpolation depends on the optimal interpolation function determined by analysis in the data domain of the data point based on the multidimensional sampling theory.
    [7" claim-type="Currently amended] The method according to claim 1,
    And providing a data-domain sampled network defined by a weight set that reflects the analysis of the unmodified data represented by the data points of the data domain.
    [8" claim-type="Currently amended] The method according to claim 1,
    The data point reflects a value in the associated first dimension that corresponds to the dimension in which the second and third dimensions are combined.
    [9" claim-type="Currently amended] An article constructed of a computer readable memory device comprising an executable manipulation data structure, the data structure comprising:
    Data points in the data domain,
    A multi-dimensional sampling module that analyzes data points within the data domain to determine a sampling architecture,
    An interrelated module that processes data points and sampling architectures to correlate data points and provide a data-domain sampled network that reflects the interrelationships of the data points, and
    And an interpolation module for interpolating a value in an associated first dimension of the data domain from an associated point defined by parameters in another dimension of the data domain.
    [10" claim-type="Currently amended] A data-domain sampled network device, comprising:
    Processor,
    A computer readable memory device operably coupled to a processor for storing an executable manipulation data structure, the data structure comprising:
    Data points stored in the data domain that reflect values in a first dimension,
    Multidimensional sampling module that analyzes data points in the data domain for second and third dimensions of the data domain to provide a sampling number corresponding to the sample number of the data points in the sub-domain comprising the second and third dimensions. , And
    Data that reflects the interrelationships between the second and third dimensions, in order to provide a domain sampled network, the data consisting of interrelation modules capable of processing data points and outputs of multidimensional sampling modules in the data domain. Domain Sampled Network Devices.
    类似技术:
    公开号 | 公开日 | 专利标题
    Arqub2017|Adaptation of reproducing kernel algorithm for solving fuzzy Fredholm–Volterra integrodifferential equations
    Oladyshkin et al.2012|Data-driven uncertainty quantification using the arbitrary polynomial chaos expansion
    Schiebinger et al.2018|Superresolution without separation
    Jost2010|The relation between evenness and diversity
    Eskelson et al.2009|The roles of nearest neighbor methods in imputing missing data in forest inventory and monitoring databases
    Fang et al.2006|Global response approximation with radial basis functions
    Holten et al.2009|Force‐directed edge bundling for graph visualization
    Lefebvre* et al.2004|Kalman filters for non-linear systems: a comparison of performance
    US8149238B2|2012-04-03|Method and system for adaptive maximum intensity projection ray casting
    Hachem et al.2011|Ontologies for the internet of things
    CA2351211C|2006-10-17|Text mining method and apparatus for extracting features of documents
    US6404445B1|2002-06-11|Method and system for modeling data
    US5949678A|1999-09-07|Method for monitoring multivariate processes
    US5455952A|1995-10-03|Method of computing based on networks of dependent objects
    Miehe et al.2016|Phase-field modelling of ductile fracture: a variational gradient-extended plasticity-damage theory and its micromorphic regularization
    US8122376B2|2012-02-21|Manipulating toolbar component dimensions
    Barton et al.2001|Resampling methods for input modeling
    AU735152B2|2001-07-05|Map builder system and method for enabling generic interfacing of an application with a display map generation process in a management system
    Anselin et al.2004|Web-based analytical tools for the exploration of spatial data
    US5983220A|1999-11-09|Supporting intuitive decision in complex multi-attributive domains using fuzzy, hierarchical expert models
    Diop et al.1994|Interpolation and numerical differentiation for observer design
    US7047131B2|2006-05-16|Apparatus and method for displaying detailed map of selected area
    Bai et al.2002|A blind approach to Hammerstein model identification
    US6961909B2|2005-11-01|System for displaying a hierarchical directory
    JP3548226B2|2004-07-28|Finite element method for image registration and morphing
    同族专利:
    公开号 | 公开日
    US6460024B1|2002-10-01|
    EP1114397A1|2001-07-11|
    WO2000016258A1|2000-03-23|
    CN1331821A|2002-01-16|
    MXPA01002728A|2002-04-08|
    AU6251999A|2000-04-03|
    JP2002525721A|2002-08-13|
    CA2343287A1|2000-03-23|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    法律状态:
    1998-09-17|Priority to US10092598P
    1998-09-17|Priority to US60/100,925
    1999-09-15|Priority to US09/398,519
    1999-09-15|Priority to US09/398,519
    1999-09-16|Application filed by 추후제출, 웨버 스테이트 유니버시티
    1999-09-16|Priority to PCT/US1999/021394
    2001-08-22|Publication of KR20010079857A
    优先权:
    申请号 | 申请日 | 专利标题
    US10092598P| true| 1998-09-17|1998-09-17|
    US60/100,925|1998-09-17|
    US09/398,519|1999-09-15|
    US09/398,519|US6460024B1|1998-09-17|1999-09-15|Data-domain sampled network|
    PCT/US1999/021394|WO2000016258A1|1998-09-17|1999-09-16|Data-domain sampled network|
    [返回顶部]