13 results about "Outer product" patented technology

The invention relates to the technical field of electric digital data processing, and discloses an ALS-based parallel tensor filling method and system, and the method comprises the steps: collecting feature information of a data set needing to be filled, building a tensor according to a preset data feature, and generating an initial factor matrix of the tensor; obtaining the number and position of tensor effective values, dividing sub-tensors according to hardware equipment performance to obtain a plurality of sub-tensors processed in parallel, and storing the sub-tensors in a COO format; a processing unit is distributed to one sub-tensor, reduction operation is carried out, and updated values of each row are assigned to the currently updated factor matrix; the three factor matrixes are alternately updated and iterated for multiple rounds, after each round of iteration is completed, the updated factor matrixes are subjected to outer product to obtain filled tensors, the loss between the front tensor and the rear tensor is calculated, the factor matrixes are alternately optimized, and missing value filling is achieved; the parallel processing efficiency is improved, and the overall calculation delay is reduced.

The invention discloses a psychological counseling intelligent recommendation method and system based on artificial intelligence, and relates to the field of psychological counseling cross technology, and the method comprises the steps: calculating a three-dimensional emotion vector according to an emotion axis space and a syntactic structure, constructing a psychological semantic tensor by using a tensor outer product, generating a user intention tensor by using a prototype weight, and carrying out a flattening operation. Obtaining a final intention vector; combining the structural potential energy deviation and the spectral distance to construct a matching factor tensor; constructing a factorization machine model, predicting a matching score, generating a prediction score matrix, screening by using a Top-K selection method, generating recommended content, and constructing a visual interface to display the recommended content. According to the method, the fine granularity and accuracy of emotional representation are improved through emotional axis space and syntactic dependency analysis, the matching precision of intention and resources is improved through comprehensive calculation of semantic gravitation intensity and spectral distance, and the expression ability of matching factor tensor is enhanced through fusion of structural potential energy deviation and spectral distance.

This invention proposes a robotic arm obstacle avoidance path planning method based on an improved ant colony algorithm, relating to the fields of robot path planning and intelligent optimization algorithms. This method endows the joint configuration space with a non-uniform metric structure induced by a metric tensor G(q), where G(q) is derived from the normalized joint inertia matrix W. I The algorithm consists of three parts: the singularity gradient outer product term and the obstacle spacing gradient outer product term. Local geodesic distances are approximated using the mean of the metric tensors at both ends of the node. All edge weights are pre-calculated and cached during the PRM graph construction phase. The ant colony uses the reciprocal of the geodesic distance as a heuristic function, drives non-uniform pheromone evaporation using the normalized value of the metric tensor trace increment, and uses the weighted sum of geodesic length cost and inertia-weighted velocity mutation penalty as the comprehensive path cost. The beetle whisker algorithm performs pre-search and completes non-uniform pheromone initialization under geodesic metrics. This method effectively improves path safety, continuity, and dynamic adaptability.

The embodiment of the invention belongs to the technical field of computers, and particularly relates to a polynomial calculation method and device and computer equipment. In the method, a first polynomial can be split into a product of a reference polynomial and a compensation exponential power, and outer product operation is performed on X first vectors corresponding to exponential powers of N input data and a second vector formed by coefficients of monomials of the same index in the reference polynomial through a matrix operation unit, and the matrix obtained by the outer product operation is accumulated to obtain the calculation result of the reference polynomial of the first polynomial corresponding to the plurality of input data, and then the calculation result is multiplied by the corresponding compensation exponent power to obtain the calculation result of the first polynomial. By the adoption of the method and device, part of the calculation process of the polynomial can be converted into matrix operation, implementation is achieved through the matrix operation unit, and the calculation efficiency of the polynomial can be improved.

The invention relates to a lightweight efficient parameter fine tuning processing method for a large model, which belongs to the technical field of natural language processing, and comprises the following steps of: converting information to be processed into a vector form through a coding layer of a pre-training model to obtain an input vector; introducing a dimension reduction matrix and a dimension raising matrix beside a query matrix and a value matrix of the pre-training model, wherein the dimension reduction matrix and the dimension raising matrix are both generated by outer accumulation of one-dimensional vectors; multiplying the input vector by the dimension reduction matrix and the dimension raising matrix in sequence to obtain a first feature vector; multiplying the input vector by an original weight matrix which is kept frozen in the pre-training model to obtain a second feature vector; according to the method, only the four groups of one-dimensional vector parameters for generating the dimension reduction matrix and the dimension raising matrix are finely adjusted, so that the problems of high calculation cost and difficulty in deployment in a resource-limited scene due to the fact that a mode of adjusting all parameters of a pre-training model is adopted in traditional full parameter fine adjustment are solved.

The present invention relates to the technical field of electronic digital data processing, and discloses a parallel tensor filling method and system based on ALS. The method collects characteristic information of a data set to be filled, establishes a tensor according to preset data characteristics, and generates an initial factor matrix of the tensor; obtains the number and position of effective values of the tensor, divides the sub-tensors according to the performance of the hardware device, obtains multiple sub-tensors for parallel processing, and stores the sub-tensors in COO format; allocates a processing unit to each sub-tensor, performs a reduction operation, and assigns the updated value of each row to the currently updated factor matrix; alternately updates three factor matrices, performs multiple rounds of iterations, and after each round of iteration, performs an outer product on the updated factor matrices to obtain a filled tensor, calculates the loss between the previous and next tensors, alternately optimizes each factor matrix, and implements missing value filling. The present invention improves parallel processing efficiency and reduces overall computing delay.

The present invention provides a method and system for singularity analysis of a multi-loop coupling mechanism based on geometric algebra, comprising: selecting a fixed platform and a movable platform of the mechanism, dividing a basic branch and a coupled branch coupled thereto in a closed loop, writing the motion spiral of the kinematic pair on each branch, and obtaining its constraint space by calculating the motion space of each branch; determining the relationship between the constraint space of the coupled branch and the basic branch, and equivalently forming a new constraint space; performing the above calculation steps on all basic branches containing a closed loop; removing redundant constraint spirals on each new branch after the closed loop is equivalent, calculating the outer product of the constraint space of all equivalent branches, obtaining the outer product coefficient, and setting the outer product coefficient to 0 to obtain the singular configuration of the mechanism. The singularity analysis method of the present invention only needs to determine the motion spiral of each branch of the multi-loop coupling mechanism to calculate the branch constraint space, and then obtain the singular configuration of the mechanism, thereby conveniently and concisely performing singularity analysis on the multi-loop coupling mechanism.

The invention provides an asynchronous mode system for executing graphic analysis in a fault-tolerant environment. The invention further discloses a system for randomizing trace approximation calculation based on an asynchronous calculation system structure. The system retrieves an adjacency matrix associated with the complex graph. The system determines a random vector based on the retrieved adjacency matrix. The system generates a matrix vector based on the adjacency matrix and the random vector. The system determines a first set of natural numbers based on a first dimension of the adjacency matrix. The system selects a subset of entries from the generated matrix vector based on the determined first set of natural numbers. The system determines a diagonal random matrix based on a sum of canonical outer products formed by the selected subset of entries. The system calculates a trace approximation of the adjacency matrix based on the determined diagonal random matrix and the selected subset of entries, and stores the calculated trace approximation of the adjacency matrix.

Adapting a pretrained model to a task. The method includes providing the pretrained model including a layer configured to map a multidimensional input depending on weights to a multidimensional output, wherein a vector includes a subset of the weights that weighs the elements of the multidimensional input for a dimension of the output of the layer; providing training data and learning at least one vector of a transformation for adapting the subset depending on the training data and the output of the model, the at least one vector having unit length, and the transformation includes an outer product of the at least one vector with the transposed at least one vector, or the at least one vector is normalized to have unit length, and the transformation includes an outer product of the normalized at least one vector with the transposed normalized at least one vector.

A method for adapting a particular pre-training model (106) to a task, where the method comprises providing the particular pre-training model (106), where the particular pre-training model (106) comprises a layer configured to map a multi-dimensional input of the layer to a multi-dimensional output of the layer depending on a weight, where a vector comprises a subset of the weight, where the subset of the weight is a subset of the weight. The subset weighs elements of the multi-dimensional input for dimensions of the output of the layer, where the method comprises: providing training data, and learning at least one vector for adapting transformations of the subset of weights depending on the training data and the output of the model (106), where the at least one vector has a unit length, the at least one vector is normalized to have a unit length, and the transform comprises an outer product of the at least one vector and a transpose of the at least one vector, or wherein the at least one vector is normalized to have a unit length, and the transform comprises an outer product of the normalized at least one vector and a transpose of the normalized at least one vector.

The invention discloses an application-specific integrated circuit and method for matrix multiplication. In some embodiments, the application-specific integrated circuit is configured to perform mathematical operations. For the matrix A and the matrix B, all product terms for determining the matrix A and the matrix B can be obtained by using the outer product (calculating all i) of each column i [vector Ai] of the matrix A and the corresponding row i [vector Bi] of the matrix B. By adding the elements of the calculated outer product, a product matrix C (i.e., A * B = C) can be assembled. Each outer product of Ai and Bi may be calculated by a series of vector-scalar products. Each vector-scalar product is calculated using selected elements of the vectors Bi and Ai as scalar. Therefore, an outer product of Ai and Bi can be generated by calculating a vector-scalar product for all elements of Ai.

A system for randomized trace approximation calculation based on asynchronous computing architecture is disclosed. The system retrieves an adjacency matrix associated with a complex graph. The system determines a random vector based on the retrieved adjacency matrix. The system generates a matrix-vector based on the adjacency matrix and the random vector. The system determines a first set of natural numbers based on the first dimension of the adjacency matrix. The system selects a subset of entries from the generated matrix-vector based on the determined first set of natural numbers. The system determines a diagonal random matrix based on a summation of canonical outer products formed by the selected subset of entries. The system calculates a trace approximation of the adjacency matrix based on the determined diagonal random matrix and the selected subset of entries and stores the calculated trace approximation of the adjacency matrix.

To reduce the number of times of access to an RF per product-sum operation.SOLUTION: For an L * N matrix A, an N * M matrix B, an L * M matrix C, and Cin, where L and M are both integers greater than or equal to 2 and N is an integer greater than or equal to 1, accumulating an outer product Ok of a k-th column of A and a k-th row of B in an array of L * M accumulators for an integer k greater than or equal to 0 and less than N; An array of l * m product-sum operators that perform an L * M * N matrix product operation C = A * B or an L * M * N matrix product-sum operation C = A * B + Cin, where either l or m is l = L or m = M, and the other is an integer of 2 ≤ l <L or 2 ≤ m <M, and the l * m product-sum operators perform accumulation of an outer product Ok in a plurality of steps.SELECTED DRAWING: Figure 17