47 results about "High dimensional" patented technology

Systems, computer programs, devices, and methods that enable LLM-based user interfaces within real-time and / or embedded devices. Providing user-specific context to a generically trained LLM may enable a variety of new usages and scenarios. For example, adaptive prompt augmentation may enable a user device to augment user-generated prompts with additional user context in the form of machine-generated prompts. In some variants, machine-generated prompts may be further refined to accommodate e.g., foundation model constraints, etc. APIs for user-specific data structures can be used to e.g., optimize for habitual behaviors, user idiosyncrasies, etc. Agentic query construction may enable a user device to operate with autonomy and decision-making capabilities, beyond prompt-response interactions. Stitching (or dreaming) may be used to identify pattern-based associations within high dimensional space (embedding vectors).

The invention relates to the technical field of metal plastic processing, and provides an integrated digital twinning large-specification bar rolling dynamic deformation distribution optimization method which comprises the following steps: constructing a digital twinning model in a large-specification bar rolling process; equipment state data, material parameters and process parameters in the rolling process are collected in real time, and the digital twin model is synchronously updated; based on the digital twinborn model, predicting strain field distribution, temperature field distribution, material metallographic structure performance and potential defect risk of the rolled piece in the current pass; with the minimum total rolling energy consumption, the highest size precision and the optimal structure uniformity as multiple objectives, deformation distribution parameters of subsequent passes are dynamically distributed through a hybrid optimization algorithm; and the optimized deformation distribution parameters are issued to a rolling mill control system to be executed, and model parameters are corrected based on online detection data feedback. In this way, the problems that in traditional rolling, due to strong experience dependence and poor dynamic working condition adaptability, efficiency is low, and many defects exist are solved.

Systems, computer programs, devices, and methods that enable LLM-based user interfaces within real-time and / or embedded devices. Providing user-specific context to a generically trained LLM may enable a variety of new usages and scenarios. For example, adaptive prompt augmentation may enable a user device to augment user-generated prompts with additional user context in the form of machine-generated prompts. In some variants, machine-generated prompts may be further refined to accommodate e.g., foundation model constraints, etc. APIs for user-specific data structures can be used to e.g., optimize for habitual behaviors, user idiosyncrasies, etc. Agentic query construction may enable a user device to operate with autonomy and decision-making capabilities, beyond prompt-response interactions. Stitching (or dreaming) may be used to identify pattern-based associations within high dimensional space (embedding vectors).

Embodiments detect anomalous activity in networks. Events may be generated based on an activity observed in a monitored network such that each event includes values associated with the activity. High dimensional event vectors may be generated by embedding based on the events and the values included in each event. Anomalous events may be determined based on detection models trained with a cluster of events associated with the high dimensional event vectors such that each anomalous event may correspond to a high dimensional event vector compared to conditions declared in the detection models and such that each anomalous event may be associated with a priority score or a confidence score. A user interface that displays a report that includes the anomalous events may be generated and arranged based on the priority score, the confidence score, a user selected preference, feedback metrics associated with the user interface, or the like.

Systems, computer programs, devices, and methods that enable LLM-based user interfaces within real-time and / or embedded devices. Providing user-specific context to a generically trained LLM may enable a variety of new usages and scenarios. For example, adaptive prompt augmentation may enable a user device to augment user-generated prompts with additional user context in the form of machine-generated prompts. In some variants, machine-generated prompts may be further refined to accommodate e.g., foundation model constraints, etc. APIs for user-specific data structures can be used to e.g., optimize for habitual behaviors, user idiosyncrasies, etc. Agentic query construction may enable a user device to operate with autonomy and decision-making capabilities, beyond prompt-response interactions. Stitching (or dreaming) may be used to identify pattern-based associations within high dimensional space (embedding vectors).

The invention belongs to the technical field of powder metallurgy, and particularly relates to a high-precision injection-molded titanium alloy workpiece and a machine learning assisted preparation method thereof.The machine learning assisted preparation method of the high-precision injection-molded titanium alloy workpiece comprises the steps that an original data set is obtained through three-dimensional mold filling numerical simulation, performing data processing on data in the original data set to obtain an expanded data set; based on the expanded data set, multiple different machine learning algorithms are adopted to construct an injection molding process parameter prediction model, and an optimal prediction model is obtained; and optimal process parameters are obtained based on the model, and a target titanium alloy workpiece is prepared. According to the method provided by the invention, three-dimensional mold filling numerical simulation and machine learning are fused, and the process parameter prediction model is constructed and optimized, so that the actual MIM process is guided, the titanium alloy workpiece with high dimensional precision can be obtained, the research and development period is greatly shortened, and the trial-manufacturing cost is reduced.

The invention discloses a milling path planning method and system for stress gradient steepest descent, and the method comprises the steps: carrying out the discretization of a residual stress field based on a lightweight bearing member, and mapping the discretized residual stress field to a milling model, so as to form a high-dimensional matrix; calculating a gradient vector and stress streamline distribution of the residual stress field, and according to the gradient vector and the stress streamline distribution, based on a minimum potential energy principle and a stress steepest descent strategy, performing milling path planning degradation to obtain an optimized baseline; and according to the high-dimensional matrix and the optimized baseline, milling path planning is converted into graph complete traversal with the graph theory as guidance, a steepest descent scheme of strain energy and rigidity is optimized, and an optimal milling path scheme is obtained. The method serves for precision machine manufacturing process design, and the problems of array geometric feature milling path planning and machining deformation control of current lightweight force bearing components are solved; and the method is particularly suitable for light-weight bearing components with array geometric characteristics and high dimensional precision.

Systems, computer programs, devices, and methods that enable LLM-based user interfaces within real-time and / or embedded devices. Providing user-specific context to a generically trained LLM may enable a variety of new usages and scenarios. For example, adaptive prompt augmentation may enable a user device to augment user-generated prompts with additional user context in the form of machine-generated prompts. In some variants, machine-generated prompts may be further refined to accommodate e.g., foundation model constraints, etc. APIs for user-specific data structures can be used to e.g., optimize for habitual behaviors, user idiosyncrasies, etc. Agentic query construction may enable a user device to operate with autonomy and decision-making capabilities, beyond prompt-response interactions. Stitching (or dreaming) may be used to identify pattern-based associations within high dimensional space (embedding vectors).

A fault diagnosis method for the RF front-end circuit of a MIMO system includes using a Synchronous Enhancement Extracting Transform (SEET) to pre-process the acquired fault signal to extract fault feature, creating a fault identification model fused by a complex field based asymmetric convolutional neural network and a complex field based multi-head attention module to assign fault feature weights, extract key feature and identify fault status. The SEET can extract fault feature components, and calculate its real field feature and imaginary field feature to obtain a real field two-dimensional matrix i and an imaginary field two-dimensional matrix q, thus an enhanced time-frequency feature is obtained. The fault identification model is used for a transform from a complex field feature space to a high dimensional space and realizing the assignment of fault feature weights, key features extraction and fault status identification.

A system and method are provided for training a neural network for controlling operation of a system having non-linear dynamics represented by partial differential equations (PDEs). The method includes collecting a digital representation of time series data indicative of an instance of a function space of the system and a measurement of a state of operation of the system. A configuration point corresponding to the solution of the PDE is generated. A neural network is trained using training data including the collected time series data and the configuration points to train parameters of the non-linear operator. The neural network has an autoencoder architecture, the autoencoder architecture comprising: an encoder to encode each instance of training data into a potential space; a non-linear operator for propagating the encoded instance into a potential space using a transformation determined by a parameter of the non-linear operator; and a decoder to decode the transformed encoded instance of the training data to minimize the hybrid loss function.

The present invention relates to a method for thermal video surveillance based on an Encoder-Decoder-induced feature pooling module. The method is explained as follows: An input thermal image that is to be processed, by a pre-trained ResNet-152 deep learning network, wherein the network comprises several convolutional layers, batch normalization layers, and a rectified linear unit (ReLU) function to extract features at low, mid and high levels, and the in-depth target features; receiving, by a feature pooling module (FPM), from the deep learning network, wherein the feature pooling module comprises of a max pooling layer, a convolutional layers, and various atrous convolutional layers for extracting the target features in higher dimensional features space at multi-scales; obtaining higher dimensional features space by a decoder network, wherein the decoder network is configured to project the higher dimensional features into image space for the generation of a probability mask.

The invention discloses a wheel forging and rolling composite forming process based on deformation distribution and a structure regulation and control method, and belongs to the technical field of wheel manufacturing. The method comprises the steps that a composite forming path of pre-forging, finish forging and finish rolling is adopted, the deformation amount of each procedure is scientifically distributed, pre-forging accounts for 70%-80%, finish forging accounts for 15%-20%, and finish rolling accounts for 5%-10%, so that forming precision and microstructure control is achieved. Meanwhile, quantitative prediction is carried out on microstructure evolution in the whole forming process by constructing a multi-scale digital simulation platform integrating a material constitutive model, a microstructure evolution model and a cellular automaton. And based on a prediction result, active and accurate regulation and control of microstructures of different parts of the wheel are realized through virtual iteration optimization of key process parameters such as temperature, deformation and strain rate, and finally, the complex-structure wheel with high dimensional precision, uniform and fine structure and excellent comprehensive performance is obtained. According to the wheel shape-character collaborative manufacturing method, shape-character collaborative manufacturing of the wheel is achieved, and the scientificity and efficiency of process design are improved.

The invention provides a digitization method for mining key elements of cultural relics, and aims to support the protection work of the cultural relics through a high-precision point cloud file generation technology. According to the method, triangular laser scanning, structured light scanning, photogrammetry and other technologies are combined, the high-dimensional-precision overall shape and contour of the cultural relic can be rapidly obtained, and meanwhile point cloud data with high texture detail precision and high color precision are obtained. The multi-technology fusion method not only improves the data processing efficiency, but also ensures the high precision of the finally generated point cloud model in size, detail and color. In addition, the method further comprises a grid model generation step based on the point cloud model, and convenience is provided for follow-up utilization and development of the cultural relic digital model. The method has advancement and practicability in technology, and is of great significance in improving the efficiency and quality of cultural relic protection and promoting inheritance and development of cultural relics.

This relationship visualization device comprises: a relationship information acquisition unit 1 that acquires relationship information representing a relationship between a central person and another person other than the central person; and a display unit 2 that displays the position of the other person in a predetermined two- or higher dimensional coordinate system determined on the basis of the relationship information.

The present invention provides a method for preparing a multi-layer modulated structure fusion modulation target, comprising: establishing a three-dimensional model of the modulation target, the three-dimensional model of the modulation target including a base layer model and multiple modulation structure layer models provided on the base layer model, the multiple modulation structure layer models being arranged in sequence and having height differences; printing the base layer according to the three-dimensional model of the modulation target; printing multiple modulation structure layers in sequence on the base layer according to the three-dimensional model of the modulation target; and obtaining a multi-layer modulated structure fusion modulation target after ultraviolet curing and plasma surface treatment. The present invention prepares partitioned, layered, and diversified modulation structures based on 3D printing, and performs plasma roughening treatment. It has the advantages of precise and controllable modulation patterns, complex and variable structures, high dimensional resolution, simple and efficient processes, and low costs, and is helpful in simulating the effects of various uneven factors on the surface of the target pellet on the fusion ignition process.

A system and method for Medical Claims Risk Score (MCRS) algorithmic underwriting includes a predictive machine learning model configured to generate underwriting decisions on electronic applications. MCRS underwriting applies word embedding modeling, such as GloVe (global vectors), to transform high dimensional MC records into single-code word vectors. These single-code word vectors are employed in regression modeling, and may include summarized embedding coordinates aggregated at the applicant level. Regression modeling uses medical claim codes data and underwriting decision data stored for historical underwriting applicants to train a random forest model to predict relative mortality risk for underwriting applicants. A risk rating may be derived from the underwriting decision data based upon standard quantitative risk ratings of a plurality of risk classes. Other inputs to the random forest model may include cohort level applicant profile data, such as applicant issue age and sex.

In some implementations, a device may obtain a training corpus, from a set of pre-processed log data, associated with an alphanumeric format. The device may encode the training corpus to obtain encoded data using a set of tokens. The device may calculate a sequence length based on a statistical parameter associated with the training corpus. The device may generate a set of input sequences and a set of target sequences based on the encoded data, where each input sequence and each target sequence has a length equal to the sequence length. The device may generate a training data set based on combining the set of input sequences and the set of target sequences. The device may train a deep neural network (DNN) using the training data set and based on one or more hyperparameters to obtain a set of embedding tensors associated with an embedding layer of the DNN.

The application discloses a kind of heterogeneous multi-view enhancement subsequence unit learning session recommendation method, specifically related to the technical field based on session recommendation, solve the interaction between single item in the prior art, lack rich global context information, it is difficult to understand the intention of user from higher dimensional angle technical problem;Its technical scheme is: multiple continuous items are regarded as a subsequence unit, and learning is carried out on the subsequence level;It explores the intention of user within a certain range by subsequence, but not just focus on the direct relationship between items;The number of items in subsequence can be dynamically adjusted, so as to explore the influence of subsequence of different lengths on recommendation performance;The application can apply subsequence unit learning user intention in multiple sessions, better capture the context information in global session.

Systems and methods are provided for simplifying the generation / application of machine learning models in a network or other deployment of elements or objects of interest. Data (which can be multi-variate, high dimensional, time-series) regarding or associated with such objects may be represented as random matrices, which can then be transformed diagonal variance matrices. Upper and lower confidence bounds can be determined with which to test similarity between the now, diagonal matrices. Based on the determined similarity or dissimilarity, one or more clusters of matrices, representative of the objects of interest, can be determined. In this way, machine learning models can be trained and developed to be operationalized for the clustered matrices (objects) rather than individual matrices (objects).

This disclosure relates generally to method to determine an optimal set of atom centered symmetry functions. One or more parameters associated with one or more atom centered symmetry functions (ACSFs) are received. An initial set of ACSFs is generated by varying the one or more parameters. A histogram with a prespecified bin size is constructed to obtain a distribution of value of each of the initial set of ACSFs. A pruned list of ACSFs is obtained based on width and maximum value of the distribution of the value of initial set of ACSFs. The pruned list of ACSFs is sorted in decreasing order of spread to obtain a sorted list of ACSFs. An optimal set of one or more shortlisted ACSFs is determined by traversing through the sorted list of ACSFs. A high dimensional neural network potential is trained based on the optimal set of one or more shortlisted ACSFs.

Groups of patrons may be discovered by measuring website and mobile site patron clickstream data in a mathematical and unsupervised way over a predetermined time and by graphically clustering the patron clickstream data using non-linear dimensionality reduction in the form of a Uniform Manifold Approximation and Projection algorithm (UMAP). The data from the UMAP may then be feed into a Density Based Spatial Clustering of Applications with Noise algorithm (DBSCAN) in order to identify a center of each cluster. Next, using the data from the UMAP and the center of each cluster from the DBSCAN, a K-Nearest Neighbor algorithm (KNN) may be applied to identify data points closest to the center of each cluster and to shade each of the data points to graphically identify each cluster of the plurality of clusters. Next, illustrate a graph on the display representative of the data points shaded following application of the KNN.

The application discloses a kind of based on layered Gaussian model's electromagnetic relay multi-fidelity transfer agent modeling method, the method uses three-layer Gaussian process model of embedded ARD kernel function as modeling method, relies on the multi-source fusion data of physical mechanism-history product data-finite element simulation-target product data, realizes the multi-fidelity small sample transfer agent model establishment of electromagnetic relay.This method introduces transfer learning, while considering the calculation limitation of finite element simulation under medium-high dimensional conditions, greatly reduces the agent model training cost of multiple electromagnetic relays with the same structure, while maintaining good calculation accuracy.The method of the application generally requires less than 10% of the ordinary model under the condition that the model accuracy and ordinary Gaussian process regression agent model are flat, and the model is more applicable.

A computer-implemented method of predicting a remaining useful life of a device or a component thereof is disclosed, the method comprising: receiving high dimensional data corresponding to operating parameters of the device or a component thereof; converting the high-dimensional data into health indicator data, wherein the health indicator data comprises low-dimensional data indicating the health state of the equipment or the components thereof; determining a temporal relationship within the health indicator data to obtain health indicator trajectory data; receiving historical maintenance data of the equipment or components thereof; determining a probability of failure of the device or a component thereof based at least in part on the historical maintenance data and the health indicator trajectory data; and predicting the remaining useful life of the device or a component thereof based at least in part on the probability of failure.

A method for high dimensional spatial analysis includes segmenting, into a plurality of segments, an image depicting a plurality of cells comprising a biological sample. Each segment of the plurality of segments may correspond to one cell of the plurality of cells. A phenotype for each cell of the plurality of cells depicted in the image may be determined based on the segmented image. The determining of the phenotype may include identifying, within the plurality of cells, a first cell type having a first phenotype and a second cell type having a second phenotype. One or more metrics, such as a colocation quotient or an Earth Mover's Distance, quantifying a co-occurrence pattern between the first cell type and the second cell type may be determined. A visual representation of the co-occurrence pattern between the first cell type and the second cell type may be generated based on the metric.

Methods and systems for determining information for a specimen are provided. One system includes a computer system and one or more components executed by the computer system. The one or more components include a pre-trained vision foundation model (VFM) configured for projecting multiple images for a specimen to high dimensional embeddings via continuous pretraining. The multiple images include an image generated for the specimen with one or more modes of an imaging system. The one or more components also include one or more additional components configured for determining information for the specimen from the high dimensional embeddings.

Methods, systems, and apparatus, including computer programs encoded on computer storage media for identification and re-identification of fish. In some implementations, first media representative of aquatic cargo is received. Second media based on the first media is generated, wherein a resolution of the second media is higher than a resolution of the first media. A cropped representation of the second media is generated. The cropped representation is provided to the machine learning model. In response to providing the cropped representation to the machine learning model, an embedding representing the cropped representation is generated using the machine learning model. The embedding is mapped to a high dimensional space. Data identifying the aquatic cargo is provided to a database, wherein the data identifying the aquatic cargo comprises an identifier of the aquatic cargo, the embedding, and a mapped region of the high dimensional space.

The application discloses a kind of based on comprehensive data analysis's software user experience evaluation method and system, it is related to human-computer interaction technical field, this method is by embedding behavior perception engine in software operation interface, collect the use frequency of user in task path, jump operation frequency, process standard step number and access user total number, and further unified normalization processing is carried out in background server, constructs normalized behavior dataset, calculates path deviation joint index PDI.In based on path deviation joint index PDI joint consideration path selection information entropy and path jump rate mean two dimensions, whether the use of user path exists concentrated deviation or process jump behavior can be effectively identified, and then whether there is forced guidance, process hidden shortcut in system path design is judged.Compared with the evaluation mode of traditional only relying on click heat map or jump rate, the present method realizes higher dimensional path behavior modeling and deviation identification ability.

Groups of patrons may be discovered by measuring website and mobile site patron clickstream data in a mathematical and unsupervised way over a predetermined time and by graphically clustering the patron clickstream data using non-linear dimensionality reduction in the form of a Uniform Manifold Approximation and Projection algorithm (UMAP). The data from the UMAP may then be fed into a Density Based Spatial Clustering of Applications with Noise algorithm (DBSCAN) in order to identify a center of each cluster. Next, using the data from the UMAP and the center of each cluster from the DBSCAN, a K-Nearest Neighbor algorithm (KNN) may be applied to identify data points closest to the center of each cluster and to shade each of the data points to graphically identify each cluster of the plurality of clusters. Next, illustrate a graph on the display representative of the data points shaded following application of the KNN.

A facility for analyzing the features of data items organized into clusters is described. The facility analyzes the data items of the clusters when the features of the data items are high dimensional and categorical with overlapping values across the clusters. The facility identifies the most distinguishable features that uniquely differentiate the clusters given the above nature of the feature space.

The application discloses a data aggregation method based on multi-modal features, comprising: collecting multi-modal data, pre-processing, extracting features according to modes and dividing into high, medium and low dimensional data, for high dimensional data, using ball tree algorithm to locate the near neighbor point; medium dimensional data is based on distribution density to dynamically adjust the neighborhood range; the low dimensional data is calculated by the Euclidean distance, and then is mapped to the low dimensional space by the aid of the local linear embedding, then traverses the low dimensional data, and the discrete data value frequency and the continuous data probability density are counted, the marginal probability is calculated by combining the information entropy, and the data aggregation weight is determined, finally, the features after dimension reduction are spliced in the order of high, medium and low levels, the probability normalization is carried out in each level block, and the aggregated comprehensive features are generated. The method realizes the aggregation of multi-modal features through multi-dimensional differentiated processing and weight calculation based on data distribution, and improves the feature complementarity and accuracy.