480 results about "Partition of unity" patented technology

A method of forecasting, allocating and scheduling in a single or multi-site skills-based contact or call center environment organized into a hierarchy of zero or more business units at a first, upper level, one or more contact types at a second, intermediate level, and one or more management units at a third, lower level. A user creates (a) a set of given contact allocations that define how contacts are distributed from a given business unit to multiple contact types, and (b) a set of given requirement allocations that define how agent requirements are distributed from a contact type to one or more management units. Agent availability by contact type is then predicted to generate agent availability data. Thereafter, forecasted contacts and forecasted agent requirements are allocated based on the given contact and requirement allocations as well as the agent availability data. Preferably, the agent availability data is predicted using a schedule simulator and is characterized by contact type for agents in the same skill group.

The invention relates to a graph-based semi-supervised high-spectral remote sensing image classification method. The method comprises the following steps: extracting the features of an input image; randomly sampling M points from an unlabeled sample, constructing a set S with L marked points, constructing a set R with the rest of the points; calculating K adjacent points of the points in the sets S and R in the set S by use of a class probability distance; constructing two sparse matrixes WSS and WSR by a linear representation method; using label propagation to obtain a label function F<*><S>, and calculating the label prediction function F<*><R> of the sample points in the set R to determine the labels of all the pixel points of the input image. According to the method, the adjacent points of the sample points can be calculated by use of the class probability distance, and the accurate classification of high-spectral images can be achieved by utilizing semi-supervised conduction, thus the calculation complexity is greatly reduced; in addition, the problem that the graph-based semi-supervised learning algorithm can not be used for large-scale data processing is solved, and the calculation efficiency can be improved by at least 20-50 times within the per unit time when the method provided by the invention is used, and the visual effects of the classified result graphs are good.

The invention discloses the hardware architecture of a binary weight convolution neural network accelerator and a calculation process thereof. The hardware architecture comprises three double-ended on-chip static random access memories which are used for buffering the binary weight of input neurons and a convolution layer, four convolution processing units capable of controlling calculation parts to complete major convolution calculation operation according to the calculation process, a feature map accumulation unit and a convolutional accumulation array. The feature map accumulation unit and the convolutional accumulation array are used for further processing the operation result of the convolution processing units to acquire a final correct output neuron value. The entire design exchanges data with an off-chip memory via a dynamic random access memory interface. In addition to the hardware architecture, the invention further provides the detailed calculation process which optimizes the hardware architecture and uses four lines of input feature map as a complete calculation unit. According to the invention, input data are reused to the greatest extent; the access of the off-chip memory is eliminated as much as possible; the power consumption of the deep binary convolution neural network calculation can be effectively reduced; a deep network is supported; and the scheme provided by the invention is a reasonable scheme which can be applied to an embedded system of visual application.

A method and apparatus for constructing one of a neuroscience-inspired artificial neural network and a neural network array comprises one of a neuroscience-inspired dynamic architecture, a dynamic artificial neural network array and a neural network array of electrodes associated with neural tissue such as a brain, the method and apparatus having a special purpose display processor. The special purpose display processor outputs a display over a period of selected reference time units to demonstrate a neural pathway from, for example, one or a plurality of input neurons through intermediate destination neurons to an output neuron in three dimensional space. The displayed neural network may comprise neurons and synapses in different colors and may be utilized, for example, to show the behavior of a neural network for classifying hand-written digits between values of 0 and 9 or recognizing vertical/horizontal lines in a grid image of lines.

An object of the present invention is to make it possible to adjust the amount of motion blur contained in a blurred image. A unit-of-processing determining portion 901 determines the unit of processing which is formed of pixel data located on a straight line. An equation generator 903 generates simultaneous equations consisting of a plurality of relational expressions based on the unit of processing. A calculator 904 generates foreground object components in which the amount of motion blur is adjusted by solving the simultaneous equations. The present invention can be applied to an image processing apparatus in which a difference between a signal detected by a sensor and the real world is taken into consideration.

The invention discloses a matrix two-dimensional bar code and coding and decoding method, which is characterized by the following: the code graph of bar code is composed of gapless arrangement ní‡m rectangle unit information blocks, wherein m and n is positive integral more than 1; n is different from or the same as m; each rectangle unit information block is composed of gapless arrangements n'í‡m' of different optical characters with two rectangle colorful blocks, wherein n' and m' is positive integral more than 1; n' is different from or the same as m'; each rectangle unit information block contains location ID character and data region; the shape of ID character displays L type or inversed L type along two adjacent edges of rectangle unit information block frame. The bar code contains the following characters: large capacity, high error-correcting rate, high pollution-proof and distortion-proof ability, multiple language editing codes, separating coding and decoding code, random scale adjustment and omnibearing recognizing, which can be applied in multiple domains.

For processing samples of a signal received via a channel represented by an impulse response of W−1 coefficients, the method comprises: determining the W roots of the Z-transform of the channel impulse response; producing an intermediate signal by equalizing the received signal by a zero-forcing method or the like based on an impulse response whose Z-transform is a Z⁻¹ polynomial of degree W-p having as roots those of the W roots which are furthest from the unit circle of the complex plane; and then obtaining estimations of the transmitted signal symbols by applying a Viterbi-type equalization method or the like based on an impulse response whose Z-transform is a Z₋₁ polynomial of degree p having as roots those of the W roots which are nearest to the unit circle.

A multi-tone CW radar (14) is used to project signals from an antenna (22) and to receive returns with the same antenna. The phase differences between the outgoing signals and the returns are analyzed to determine the existence of motion and the range to a moving object (10). A model is made which has range as its major parameter. A waveform associated with the phase difference between outgoing signals and returns for one of the tones is compared to templates produced by the model to determine which has a range that most closely matches. By varying the range parameters, when a match is detected the range to the object can be obtained even if its motion is pseudorandom. If the range is measured with multiple units it is possible to measure the location of the object. This can be done assuming a grid and algorithmically combining the ranges from the units.

A unit cell for a lattice structure includes eight unit trusses disposed at vertices of the unit cell. A single unit truss is disposed at a centroid of the unit cell. Each of the nine unit trusses includes fourteen struts. Lattice structures are commonly used to connect various loads within a volume of space. Most such structures, however, have a rigid definition for their topology, and are unable to conform to shape or load directions. Additionally, conventional lattice structures are homogeneous, having dimensions and properties that are consistent throughout. These constraints, generally imposed for ease of manufacturing and assembly, prevent the development of highly robust and efficient structures, and limit the potential for multi-functional applications.

A method for the supervised teaching of a recurrent neutral network (RNN) is disclosed. A typical embodiment of the method utilizes a large (50 units or more), randomly initialized RNN with a globally stable dynamics. During the training period, the output units of this RNN are teacher-forced to follow the desired output signal. During this period, activations from all hidden units are recorded. At the end of the teaching period, these recorded data are used as input for a method which computes new weights of those connections that feed into the output units. The method is distinguished from existing training methods for RNNs through the following characteristics: (1) Only the weights of connections to output units are changed by learning—existing methods for teaching recurrent networks adjust all network weights. (2) The internal dynamics of large networks are used as a “reservoir” of dynamical components which are not changed, but only newly combined by the learning procedure—existing methods use small networks, whose internal dynamics are themselves completely re-shaped through learning.

The invention relates to a high-efficiency Latin hypercube experimental design method, belonging to the technical field of engineering optimization designs. For realizing the purpose that users extract m experimental design points from an n-dimensional design space in actual engineering, the method comprises the following steps of: presetting the first experimental design point by establishing a space hypercube; determining the second experimental design point in a second unit hypercube by utilizing the first experimental design point; determining the sth experimental design point in the sth unit hypercube by utilizing the obtained s-1 (s is equal to 3, 4, 5,......, m-1) experimental design points until the (m-1) experimental design points are determined; and uniquely determining the mth point meeting the design condition in the mth unit hypercube. The design method can obtain the experimental design points meeting the characteristic requirements of space uniformity and projection uniformity without an optimization process, thereby having high sampling efficiency; and the obtained experimental design points have good properties and are suitable for the optimization fields of any engineering designs and the like with huge operation.

A method of cubic mapping with texturing is described. Neighboring pixels on an object are mapped to adjacent faces of the cube, but these adjacent faces do not guarantee continuity in the texture mip-map associated with each face. Therefore, the u and v texture map coordinates are adjusted after mapping to adjacent faces to make a continuity adjustment that insures that the LOD for the texture mip-map is the same for each adjacent face. The continuity adjustment includes either switching the u coordinate with the v coordinate or negating one of the coordinates or both. Additionally, if the u and v coordinates are normalized, the normalization may be compensated by adding or subtracting unity from the adjusted coordinate. After the continuity adjustment is made an approximation to the derivative is computed and used to determine the LOD for the mip-map. Texturing can then proceed using the LOD.

The present invention relates to building base plates assembled to make two or three dimensional pyramids. Eight building base plates (1A˜1H) can be arranged as a rectangular plane (50) with seven cubic units in longitudinal direction, five cubic units in transversal direction; or assembled to build a five-layer (61˜65) pyramid (60) on a triangular frame each side arrayed five cubic units, a first layer arrayed fifteen cubic units. Furthermore, four building base plates (1I˜1L) with different shapes added to the eight building base plates (1A˜1H), totally fifty six cubic units of those twelve building base plates make a rectangular shape (70) with eight cubic units in longitudinal direction, seven cubic units in transversal direction, or assembled to build a six-layer (81˜86) pyramid (80) on a triangular frame, each side arrayed six cubic units, a first layer of the pyramid arrayed twenty one cubic units.

A distributed software integration service method for open robot development based on the Internet is disclosed, which makes it possible to manufacture a user-oriented robot through combination of independent heterogeneous modules. The invention involves a robot development procedure and a new robot development environment/tool for providing user-oriented services and for integrated operating of distributed software installed in the modules over the Internet. The robot development procedure includes three independent specialized development stages, i.e., a platform development stage, a module development stage, and a user-oriented robot development/user service development stage. The invention makes it possible to mass-produce autonomous robots in units of interoperable functional modules. It is also possible to meet various demands of consumers, achieve specialization, and accelerate technology development since the development procedures are specialized in an independent manner and are suitable for manufacturing a wide variety of robot products in small quantities.

The invention provides a method of simulating behavior of a flow interacting with an object. The invention improves the accuracy of the approximation of the spatial p^th derivative D_p, p≧1, and therefore reduces the cost of the simulation. It includes in the approximation values of the parameter at points which do not lay on grid lines passing through the point of computation P, and by using these values to optimize the approximation. The use of these additional points depends on a preferential direction, such as determined by the advection direction of the flow. The points used in the approximation extend beyond a unit cube on the computational grid. The simulation is in three or more dimensions. The numerical simulation produces output which can be used in the design or optimization of the object which interacts with the flow. In certain cases, it is the output by itself which is important.

The invention relates to a lightweight BIM big data online visualization method and system. The method includes the following steps that lightweight analysis based on semantic guidance is conducted on a BIM scenario component set to obtain a lightweight component base; the interior structure of a BIM building is automatically partitioned to generate multiple relatively independent spaces and a relation structure chart of the relatively independent spaces; a multilayer data index structure with multi-granularity decomposition is established according to the relation structure chart and the component set in each space; the lightweight component base is called according to the multilayer data index structure to achieve progressive visualization of the BIM building, and in the visualization process, with each space as the unit, data in the next space relative to the current space is cached and data in the last space is deleted. Compared with the prior art, the method and system have the advantages of automation, lightweight, fine granularity and the like.