14 results about "Vector process" patented technology

Quantizing tensors and vectors processed within a neural network reduces power consumption and may accelerate processing. Quantization reduces the number of bits used to represent a value, where decreasing the number of bits used can decrease the accuracy of computations that use the value. Ideally, quantization is performed without reducing accuracy. Quantization-aware training (QAT) is performed by dynamically quantizing tensors (weights and activations) using optimal clipping scalars. “Optimal” in that the mean squared error (MSE) of the quantized operation is minimized and the clipping scalars define the degree or amount of quantization for various tensors of the operation. Conventional techniques that quantize tensors during training suffer from high amounts of noise (error). Other techniques compute the clipping scalars offline through a brute force search to provide high accuracy. In contrast, the optimal clipping scalars can be computed online and provide the same accuracy as the clipping scalars computed offline.

The present application provides a case analysis method and system based on deep learning. The method includes: obtaining a target data set containing image data and text data; extracting features from the image data in the target data set based on a pre-trained convolutional neural network; inputting the extracted feature vectors into a pre-trained natural language processing model for analyzing the text data in the target data set; generating new feature vectors by linear interpolation fusion based on the image feature vectors processed by the convolutional neural network and the text feature vectors processed by the natural language processing model; expanding the target data set based on the generated new feature vectors, and training a deep learning model based on the expanded target data set to obtain a trained deep learning model; inputting the case to be analyzed as input into the trained deep learning model, and obtaining the case analysis results output by the trained deep learning model.

This invention provides a dual-stream prefetch hardware system and vector operation method for vector processing. The system includes a computational path for performing regular vector operations in response to vector operation instructions, a computational path for performing streaming vector operations in response to predefined streaming operation instructions, and a storage unit. The system includes a streaming instruction control unit and a vector streaming processing core. The streaming instruction control unit is used to identify predefined streaming operation instructions and is configured to close the computational path for regular vector operations when a predefined streaming operation instruction is detected to be fetched. The vector streaming processing core is used to perform streaming vector operations in response to predefined streaming operation instructions. This solution achieves hardware decoupling of dense layout operations and non-dense sparse streaming operations by building dual independent computational paths for regular vector operations and streaming vector operations, thereby effectively improving the overall operation efficiency and real-time performance.

The invention discloses a hierarchical bypass forwarding method and system for a vector processor, belongs to the technical field of vector processors, and aims to solve the technical problem of how to overcome the defect of low efficiency caused by read-after-write (RAW) data risk in vector operation of the vector processor, effectively reduce pause caused by VRF access delay and improve the reliability of the vector processor. According to the technical scheme, in each vector processing channel of the vector processor, before a front instruction result is written back to a local vector register file of each vector processing channel, a two-stage data forwarding architecture comprising an in-channel bypass unit and a cross-channel bypass network is established by adopting a layered result forwarding network; a front instruction result can be directly transmitted to a subsequent instruction through a two-stage data forwarding architecture comprising an in-channel bypass unit and a cross-channel bypass network; wherein each vector processing channel is internally provided with an in-channel bypass unit.

The present invention discloses a small sample process monitoring method suitable for a multi-working condition environment, which is used to solve the problem of small sample fault detection under different working conditions. The present invention embeds the prior knowledge of the working condition into a model-independent meta-learning framework to improve the learning ability. A meta-learning framework is used to construct multiple small sample tasks, and a convolutional neural network is used to extract features from the samples. An embedding vector representing the working condition mode is constructed, and a sample attention mechanism is introduced to ensure that in the process of generating the embedding vector, samples containing more working condition mode information have a higher weight. Multiple conditional layers are introduced into the basic classifier of meta-learning, and a feature linear modulation method is used to generate modulation parameters by calculating the embedding vector, and then the parameters of the basic classifier network are adjusted point by point, referring to the prior information of the working condition.

The invention discloses a near-fault multi-point multi-dimensional fully non-stationary seismic oscillation dimension reduction simulation method and system, and belongs to the technical field of seismic engineering and disaster prevention and reduction. The method provided by the invention comprises the following steps: firstly, screening near-fault strong earthquake records containing horizontal and vertical velocity pulses, fitting the extracted pulses by using a Gabor wavelet model, identifying low-frequency key parameters and establishing probability distribution; a high-frequency component is simulated by a 1D-nV non-stationary random vector process, and dimensionality reduction is carried out in combination with a random orthogonal function; and generating a representative point set through number theory point selection and equal probability inverse transformation, substituting the representative point set into the model to generate high and low frequency speed time histories, and superposing the high and low frequency speed time histories to obtain a seismic oscillation sample with complete probability information. The system comprises a data screening module, a parameter identification module and the like. According to the method, samples meeting the precision are generated by using extremely few random variables, the simulation efficiency is improved, and the method can be combined with the probability density evolution theory for engineering structure dynamic response and anti-seismic reliability analysis.

Disclosed implementations for providing a definition of probability flow between probability distributions. In an example implementation, a prompt is received from a computing device. A generative model of a vector process is conditioned based on the prompt, the generative model defined by a plurality of probability distributions of the vector process and employing a definition of a velocity field over a time interval. A vector sequence is generated with the generative model, wherein the vector sequence is an instantiation of the vector process.

The invention relates to a vector data processing method and device and electronic equipment, and the method comprises the steps: carrying out the clustering processing of input vector data through a CPU in a server, and constructing a distance query table according to the clustering result of the input vector data; sending the vector processing task based on the distance lookup table to a persistent PIM kernel of memory computing hardware PIM through a message queue; the persistent PIM kernel comprises a plurality of processing units (PU); for each processing unit (PU), receiving query distribution data of a vector processing task through a heat transmission mechanism, and based on a bus ownership switching mechanism, carrying out vector distance calculation by utilizing vector data stored in a main memory (MRAM) to obtain a distance calculation result; and through the persistent PIM kernel, the distance calculation result after the PU sorting of each processing unit is transmitted back to the CPU for merging processing, and a vector data processing result of the vector processing task is obtained. By adopting the method, the utilization rate of the PU resources and the vector data retrieval efficiency can be effectively improved.

The present invention discloses a method for measuring the interrupt duration of an embedded system with high precision, and relates to the field of computer technology. The method comprises the following steps: S1: specifying an exception vector number corresponding to an interrupt exception type; S2: when an interrupt exception occurs, the system jumps to the system's general exception entry function; S3: the general exception entry function determines the type of exception based on the exception type value; S4: in the interrupt entry function, calling an interrupt vector processing macro function; S5: at the entry of the interrupt vector processing macro function, first executing an interrupt duration measurement start macro function to obtain an interrupt input timestamp, then calling an interrupt processing function to obtain an interrupt output timestamp, and finally, at the exit of the interrupt vector processing macro function, executing an interrupt duration measurement end timing and data statistics macro function; S6: after the interrupt processing is completed, returning to the system state. The present invention is a non-invasive performance analysis method that can fully utilize hardware resources and improve debugging efficiency without modifying the program function flow.

The invention discloses a JSON process library-driven intelligent retrieval enhanced verbal skill generation method and system, and belongs to the technical field of artificial intelligence, and the method comprises the steps: constructing a JSON format standard verbal skill library and a vector process library based on sales verbal skill, and carrying out vectorization processing on verbal skill texts. A query statement is vectorized and matched with a verbal skill vector, and a JSON text block with the highest similarity is obtained. The method comprises the following steps: performing semantic extraction by utilizing a large model, generating an answer text of a current process, dividing the process into a'large process' serving as a main step of a sales process and a'small process' serving as a secondary step of the sales process, and dynamically updating answer content by combining the current answer text and a JSON text block of a next process. According to the method, the verbal skill selling flexibility of salesmen can be enhanced, and the updating efficiency and retrieval efficiency of the standard verbal skill library are improved.

A framework for interpreting machine learning models is proposed that utilizes interpretability methods to determine the contribution of groups of input variables to the output of the model. Input variables are grouped based on dependencies with other input variables. The groups are identified by processing a training data set with a clustering algorithm. Once the groups of input variables are defined, scores related to each group of input variables for a given instance of the input vector processed by the model are calculated according to one or more algorithms. The algorithms can utilize group Partial Dependence Plot (PDP) values, Shapley Additive Explanations (SHAP) values, and Banzhaf values, and their extensions among others, and a score for each group can be calculated for a given instance of an input vector per group. These scores can then be sorted, ranked, and then combined into one hybrid ranking.

The embodiments of this specification provide a data normalization processing method and device based on multi-party secure computation, which are applicable to the process of securely determining the normalization vector of an m-dimensional first vector x stored in n data parties in a shared manner based on the normalization exponential function softmax. The basic idea includes: setting the initial normalization vector as the mean normalization vector of m-dimensional unit vectors, and iteratively correcting it through multiple update rounds for the initial normalization vector to approximate the normalization result of the normalization exponential function. In a single update round, each data party determines the offset of the current update round based on secure multiplication, and during the iteration process, the calculation results of each item are kept in a shared form. In this way, softmax can be approximated more accurately during the multi-party secure computation process, thereby improving the accuracy of softmax data normalization.

The invention relates to a large language model and vector processing off-line knowledge base integration system and method, and the system comprises an application program interface module which is used for responding to a document import request of a user, splitting a document based on a title level or semantic statistical characteristics, and calling a vectorization service to carry out vectorization storage on the split document; and responding to a chat request of the user, forwarding user query to the service management module, and returning a response generated by the large language model, the service management module is used for responding to the request of the application program interface module and automatically routing the request to chat service, vectorization service or reordering service; the chat service, the vectorization service or the reordering service is realized by calling a preset large language model, a preset vector model or a preset reordering model. According to the method, knowledge base management and chat question and answer are completed through API interface service, integration of a large language model and vector processing is achieved, and instant use of common users is achieved.

A framework for interpreting machine learning models is proposed that utilizes interpretability methods to determine the contribution of groups of input variables to the output of the model. Input variables are grouped based on dependencies with other input variables. The groups are identified by processing a training data set with a clustering algorithm. Once the groups of input variables are defined, scores related to each group of input variables for a given instance of the input vector processed by the model are calculated according to one or more algorithms. The algorithms can utilize group Partial Dependence Plot (PDP) values, Shapley Additive Explanations (SHAP) values, and Banzhaf values, and their extensions among others, and a score for each group can be calculated for a given instance of an input vector per group. These scores can then be sorted, ranked, and then combined into one hybrid ranking.