65 results about "Cognitive computing" patented technology

A system and method for persistent cognitive computation with temporally synchronized multimodal processing implements a geometric approach to artificial intelligence through typed latent entities within a dynamic manifold substrate. The system maintains a latent manifold incorporating heterogeneous data modalities where local curvature reflects semantic density and typed entities are stratified according to structural properties. Temporal synchronization coordinates asynchronous multimodal data streams through generation of temporal alignment fields within the manifold that preserve semantic coherence across modal boundaries. Type-aware geometric operations enforce operation legality based on entity type and local manifold geometry, enabling structured recombination, compression, and traversal while preventing semantic distortion. The system executes synchronized manifold reorganization during idle periods through coordinated optimization operations including perturbation analysis and topological surgery. This architecture enables persistent memory through geometric encoding where frequently accessed concepts develop high-curvature regions and cognitive patterns emerge from usage-based manifold evolution.

The invention discloses intelligent knowledge service electronic equipment for information retrieval, and the equipment comprises a data management layer which is responsible for providing original data for a cognitive calculation layer, and carrying out the closed-loop optimization through continuously monitoring the feedback of a man-machine interaction layer; the cognitive calculation layer is responsible for converting original data into a semantic tag system and a dynamic knowledge graph, applying BERT + GNN hybrid analysis, BERT processing text context and GNN analysis entity relationship in the knowledge graph, and dynamically adjusting a reasoning path according to user field preference by combining an SWRL rule engine and reinforcement learning; the service arrangement layer is responsible for extracting data from the multi-modal data pool, solidifying and arranging a complex query stream to perform logic decoupling and efficient multiplexing, and providing a visual interface to support non-technical personnel to configure a new scene; and the man-machine interaction layer is used for constructing a natural interaction bridge through the structured answers and the visual maps output by the service arrangement layer, carrying out personalized interface adaptation based on historical dialogues, and establishing a user feedback closed loop.

Systems and methods are provided for training an artificial intelligence system and generating audible content for output. The method utilizing a system including at least an application plane layer, a control plane layer including a cognitive computing unit, the cognitive computing unit using at least machine learning for training of the cognitive computing unit, a training input to the system including an input for receiving content for training during the machine learning, and a data plane layer, the data plane layer including an input interface to receive and store data input content from one or more data sources other than the control plane layer, the data input content being subject to transformation into audible content for output. Data input content information is used in synthesizing audible output content at least in part by transforming the data input content into the audible output content.

The present invention relates to a cognitive computing-assisted clinical decision support system designed to enable personalized neurological rehabilitation. The system acquires structured user data across clinical, anatomical, radiological, etiological, pathological, and rehabilitation domains to create individualized profiles. These profiles are mapped against a repository of historical cases using analog matching and similarity scoring to generate stratified, evidence-based rehabilitation recommendations. Real-time monitoring of rehabilitation progress is performed using global recovery and function outcome indicators, allowing for dynamic adjustment of treatment plans. Clinician intervention modules ensure safety, interpretability, and context-aware customization. The system incorporates a continuous feedback mechanism to refine future predictions and recommendations, making it increasingly adaptive over time. The invention improves rehabilitation outcome prediction accuracy, reduces recovery variability, and optimizes functional outcomes by transforming static rehabilitation models into intelligent, responsive, and personalized care pathways.

The invention discloses a multi-parameter cooperative control method for non-excavation repair of a drainage pipe network in urban updating, and relates to the technical field of municipal engineering, and the method comprises the steps: firstly constructing a quantum entanglement sensing pipe network digital twinborn model, fusing multi-source data, monitoring micro-strain in real time through a nano sensor, deploying a self-organizing intelligent node and terahertz imaging equipment, and constructing a quantum entanglement sensing pipe network digital twinborn model; multi-dimensional features are extracted, repair parameters are dynamically regulated and controlled through technologies such as magnetofluid plugging and multi-robot cooperation based on digital twinning virtual simulation, finally, the repair effect is evaluated and the scheme is optimized in combination with cognitive calculation and microbial self-healing technologies, and full-process intelligent control is achieved. The intelligent level of non-excavation repair of the drainage pipe network is improved, the problem that traditional monitoring is one-sided is solved through high-precision data fusion, precise construction control is achieved through cooperation of magnetofluid leaking stoppage and the robot, durability is enhanced through the microorganism self-healing technology, repair quality and efficiency are improved, and urban updating and sustainable development are promoted.

The invention discloses a heterogeneous feature fusion-based multi-task traffic situation cognitive calculation method, and belongs to the field of intelligent traffic. The method specifically comprises the following steps: firstly, preprocessing multi-source traffic observation data to obtain normalized traffic flow, OD raster data and traffic state data; then, for a current frame, carrying out dynamic space-time convolution and diffusion diagram convolution on traffic flow and OD raster data to obtain node-level space-time features; meanwhile, carrying out feature extraction on the traffic state data by utilizing 3D convolution to obtain grid-level spatial-temporal features; and then, inputting the features and the common features into a heterogeneous cross attention fusion module to obtain a unified fusion feature h of the current frame, performing multi-task prediction, and respectively obtaining flow, OD and traffic state prediction results of future K steps. And finally, inputting the prediction result of each task and a true value to carry out multi-task loss calculation, carrying out parameter updating and completing model training. According to the invention, multi-task prediction can be realized by using multiple prediction heads.

The invention discloses a complex information chain modeling and self-adaptive reconstruction method based on cognitive calculation. The method comprises the steps that an information chain topology model is constructed according to nodes and links, and dynamic state parameters are configured; performing network cognitive evaluation based on the topology model and the state parameters, and generating a node efficiency index and a link reliability index; when a link failure is detected, determining an alternative path in the topology in which the failed link is removed based on the index; and according to the alternative path, reconstructing a topology model and updating parameters, and triggering new cognitive evaluation to form an adaptive closed loop. According to the method, the dynamic cognition and vulnerability analysis of the complex information chain can be realized, the adaptive reconstruction can be quickly completed when the link fails, and the robustness and task guarantee capability of the information chain in a dynamic confrontation environment can be remarkably improved.

The invention relates to the technical field of automatic driving, and discloses a Robotaxi-oriented VLA-world model fusion automatic driving system and a closed-loop training platform, and the system comprises an environment perception coding layer which is used for generating space-time aligned environment feature vectors; the mixed cognitive calculation core layer integrates a reactive sub-network, a judicial sub-network, a cognitive uncertainty evaluator and an anti-fact query interface; wherein the reactive sub-network is used for generating future trajectory distribution, and the judicial sub-network is used for calling the reactive sub-network to perform future state simulation through an anti-fact query interface when a trajectory distribution entropy value exceeds a threshold value, and executing thinking chain reasoning based on a simulation result to generate a control instruction. According to the method, deep closed loop of reactive prediction and judicial reasoning is realized through an anti-fact query mechanism, and the problems of cognitive splitting and reasoning short view existing when a traditional modular system processes a long-tail scene are solved.

The invention relates to the technical field of education and teaching. The invention provides a classroom teaching dynamic feedback and evaluation system integrated with multi-modal emotion calculation. The system comprises a multi-modal data acquisition module used for acquiring multi-source data in a classroom teaching environment and sending the multi-source data to an emotion-cognitive calculation center; the emotion-cognition calculation center is used for carrying out fusion analysis on the multi-source data, identifying the emotion state and the cognition state of the student, carrying out attribution analysis through an emotion-cognition coupling model, generating a coupling state label for describing the specific reason of the learning dilemma, and sending the coupling state label to the dynamic intervention engine; the dynamic intervention engine is used for matching and generating individual-level, group-level and system-level teaching intervention instructions from the teaching strategy knowledge base according to the received coupling state labels; and the visual feedback and evaluation module is connected with the emotion-cognition calculation center and the dynamic intervention engine, and is used for providing real-time classroom state visual display for the teacher and pushing personalized learning suggestions to the student terminal.

The invention provides a multi-agent collaborative cognitive calculation method based on point cloud feature marking, and belongs to the field of Internet of Vehicles. The method specifically comprises the following steps: firstly, inputting original point cloud data collected by a sensor of a vehicle intelligent agent into a point cloud feature mark generator, converting the original point cloud data into a one-dimensional point cloud feature mark sequence semantic perception dynamic encoder for encoding a mark sequence, and generating a feature sequence; packaging the feature sequence, the space coordinates and the pose information of the self-vehicle intelligent body into a message data packet; converting the feature mark coordinate space of the neighbor intelligent body into a coordinate system with the self-vehicle intelligent body as the center by a point cloud mark aggregation module, and generating a unified sequence; and the semantic perception dynamic fusion module carries out global context modeling and dynamic fusion on the unified sequence, corrects feature mark position deviation, generates a refined sequence, and inputs the refined sequence into a downstream task network for generating final prediction. According to the method, the perception robustness is obviously improved.

The invention provides an intention-driven semantic evolution mechanism based on a tensor semantic field and an application system thereof, and relates to the technical field of artificial intelligence, semantic networks and cognitive computing. According to the method, five types of semantic primitives including data, information, knowledge, intelligence and intention are expressed as high-order tensor nodes, a semantic tensor field network is constructed, and dynamic evolution and intention driving of semantics are achieved. The system comprises a multi-scale semantic aggregation mechanism, an intention weight diffusion algorithm, a semantic tensor evolution operator and a white box interpretation interface, supports full-link semantic processing from original data to high-level wisdom to intention constraint, and overcomes the defects in semantic representation and evolution, intention fusion and system interpretability in the prior art. And the generative AI system has stronger intention perception, semantic self-optimization and process transparency capabilities, and is suitable for applications such as a semantic perception large model platform, an AI cognitive map system and an interpretable language generator.

In one aspect, the inventions include a system for control of a software defined computer network state system. First, an application plane layer is adapted to receive instructions regarding operation of the state system. Preferably, the application plane layer is coupled to an application plane layer interface. Second, a control plane layer includes an adaptive control unit, such as a cognitive computing unit, an artificial intelligence unit or a machine-learning unit. Third, a data plane layer includes an input interface to receive data input from one or more data sources. A title transfer network element is provided to transfer digital assets via a blockchain. The system may use domain transformations and difference engines.

The invention relates to a neural electrophysiological signal analysis technology in the field of brain cognition calculation, in particular to a cross-individual EEG driving fatigue detection method based on multi-feature comparative learning. The method comprises the steps that PSD features and DE features are obtained based on electroencephalogram signal sample data; extracting a shallow PSD semantic feature, a shallow DE semantic feature, a deep PSD semantic feature and a deep DE semantic feature; obtaining a PSD prediction classification label and a DE prediction classification label based on the deep semantic features; defining a positive sample pair, a negative sample pair and a semi-positive sample pair, and constructing a contrast loss function; splicing the shallow semantic features to obtain a preliminary fusion representation, and obtaining a fusion feature vector based on the preliminary fusion representation, the deep PSD semantic features and the deep DE semantic features; and inputting the fused feature vector into a multi-feature decision maker to obtain a fatigue classification label of each sample. According to the method, the adaptability of the model to individual differences is improved, and the generalization ability of the model and the detection robustness are enhanced.

A system and method for implementing persistent cognitive computation through geometric representation of thought in a dynamic latent manifold. The system encodes inputs into a curved space characterized by time-evolving metric tensors, compression pressure fields derived from Ricci curvature, and goal potential fields that shape attention flow. Cognition occurs through geodesic traversal of this manifold, with attention following paths that minimize cognitive action while balancing semantic density and goal relevance. A Cognitive Dynamics Engine maintains manifold geometry, computing optimal trajectories and managing thought bundle operations including consolidation, expansion, and higher-order abstraction. During idle periods, autonomous dreaming processes reorganize the manifold through perturbation, recombination, and topological surgery. This architecture enables persistent memory through geometric encoding, where frequently accessed concepts develop high-curvature regions and cognitive shortcuts emerge from usage patterns, transforming artificial intelligence from stateless computation to structured motion through shaped memory space.

The invention discloses a lung cancer high-risk group screening questionnaire cognitive calculation correction method and system. The method comprises the following steps: acquiring questionnaire text data of a subject; constructing a BERT-QA model based on the clinical term knowledge graph, analyzing questionnaire text semantics, and extracting structured risk factor data; mapping the risk factor data to a topological space, and constructing a risk topological representation; calculating a dynamic weight coefficient of each risk factor based on an information entropy principle, and generating a risk assessment value of entropy weight optimization; carrying out calibration in combination with regional pollution map data; according to the method, deep learning, topology and fuzzy logic technologies are fused, intelligent analysis of questionnaire texts and dynamic weight calculation of risk factors are realized, environmental pollution factors are effectively integrated, the screening accuracy and individuation level are improved, and the method is suitable for large-scale popularization and application. The method provides technical support for accurate identification and early intervention of lung cancer high-risk groups, and has important clinical application value.

A system and method for implementing persistent cognitive computation through geometric representation augmented with holonomy-based experiential memory. The system encodes inputs into a curved latent manifold and maintains bounded sets of holonomy descriptors at each location, enabling two-component cognitive states comprising position and experiential context. Cognition occurs through holonomy-sensitive traversal where paths depend jointly on geometric structure and accumulated path-dependent constraints. Holonomy generators are created during traversal from prediction errors and constraint encounters, composed into consolidated descriptors, and undergo lifecycle management including reinforcement, decay, and irreversible export to residual constraint regions. This architecture escapes location-only representations by distinguishing cognitive states that occupy identical semantic positions but arise through different experiential histories. The system supports counterfactual reasoning through holonomy switching at fixed locations and preserves semantic memory as compressed transport deformation rather than stored trajectories, enabling scalable experiential learning where repeated patterns strengthen constraints while capacity remains bounded.

The application relates to the technical field of automatic driving and discloses a VLA-world model fusion automatic driving system and a closed-loop training platform for Robotaxi, which comprises an environment perception coding layer used for generating a space-time aligned environment feature vector; a hybrid cognitive computing core layer integrating a reactive subnetwork, a prudent subnetwork, a cognitive uncertainty evaluator and a counterfactual query interface; wherein the reactive subnetwork is used for generating a future trajectory distribution, the prudent subnetwork is used for simulating a future state through the counterfactual query interface to call the reactive subnetwork when the trajectory distribution entropy value exceeds a threshold value, and a control instruction is generated through thought chain reasoning based on a simulation result, the application realizes a deep closed loop of reactive prediction and prudent reasoning through a counterfactual query mechanism, and solves the problems of cognitive fragmentation and reasoning short-sightedness of a traditional modular system when processing long-tail scenes.

The application discloses a multi-modal false news detection method and device, and relates to the field of multi-modal cognitive computing. The method is used for providing in-depth mining and analysis of news information, realizing semantic enhancement of multi-modal information, and mining and application of news sentiment features. According to the BERT model, the first image caption information is subjected to feature coding to obtain a first caption representation and a first empirical representation. A first text representation is subjected to a self-attention mechanism, a full connection layer and averaging to obtain a sentiment representation of comprehensive analysis of the first text representation by all experts and a sentiment tendency evaluation value of comprehensive analysis of the first text representation by all experts. According to the attention weights of a plurality of different sentiment news processors and the first news processing features, news aggregation features for true or false classification are obtained. According to the news aggregation features for true or false classification and a false news detector, a prediction label of the first news is obtained.

This invention discloses a hierarchical continuous attractor network based on intrinsic time constant gradients, belonging to the fields of artificial neural networks and cognitive computing. The network includes a spatial input extraction module, multiple continuous attractor layers, a spatial projection module corresponding to each layer, and a wave packet decoding module. Each layer has a preset and fixed time constant during initialization, and the time constants of each layer monotonically increase from the bottom layer to the top layer. The same external spatial input is projected to each layer in parallel. The bottom layer tracks instantaneous spatial changes with a shorter relaxation time, while the top layer accumulates slowly varying spatial statistics with a longer relaxation time. The cyclic connection weights of each layer are initialized with translation-invariant functions to form a continuous attractor manifold, and are fine-tuned online by superimposing Hebbian enhancement and weight decay. This invention can form multi-timescale spatial representations, improving the stability and precision of representations in high-frequency spatial regions.

The embodiment of the invention provides a risk prevention and control rule configuration method and device, equipment, a medium and a program product. The method comprises the steps of obtaining voice data input by a user in response to a voice input operation of the user on a configuration page of a target financial service of a bank system; the voice data is used for describing configuration requirements related to risk prevention and control rules of the target financial service for the user; acquiring configuration information related to configuration items required by the configuration page from the voice data based on the related information of the configuration page; the related information is used for determining the context of the configuration page; filling the corresponding position of the configuration page with the configuration information; and in response to a determination operation of the user on the configuration page based on the configuration information, constructing a risk prevention and control rule of the target financial service based on the configuration information. The method is used for improving the efficiency of cognitive calculation decision system / platform configuration.

The invention relates to the field of cognitive computing, and discloses a computing co-processing hardware system and method based on user personality modeling, which comprises the following steps: aggregating records by taking task instances as boundaries, and executing time sequence correction on the aggregated records according to an operation triggering precedence relationship and a context dependency relationship; splitting the behavior data set according to the task stage switching points, and mapping the split behavior data to an operation type space; user cognitive feature description data is injected into a configuration entry of an information processing flow, and calling conditions, execution sequences and parallel relationships of information analysis nodes, result splicing nodes and intermediate data resident nodes are combined and set; continuously recording confirmation, revision and cross-stage calling behaviors of a user on a stage result; and according to the adjusted information processing node participation set, performing structured splitting and sequential recombination on the information data flow entering the system. The method has the advantage of improving the information matching accuracy.

In order to solve the problem of event prediction reasoning, the invention provides an enhanced path reasoning algorithm based on a fused context script event graph. By analyzing the limitation of an existing event prediction reasoning algorithm, an event prediction reasoning problem is modeled as a Markov decision process (MDP), and a reinforcement learning framework including a state space, an action space, a strategy function, a transfer function and a reward mechanism is constructed. In the aspect of algorithm design, design implementation of the whole algorithm is completed through a graph structure learning module, an action scoring module, a random strategy generation module, a reward calculation module and a strategy optimization module. According to the model, training and test verification are carried out on an MCNC and a self-built building construction accident report data set, the result shows that the model is superior to other existing prediction models in a downstream event prediction reasoning task, the accuracy rate is improved by 2.61% compared with an optimal RoBERTa-RF model, the result has statistical significance, and the method is suitable for popularization and application. According to the method, a new methodology framework is provided for an event prediction task, and a new path is opened up for cognitive calculation research of an event atlas through an interpretable path reasoning mechanism.

The invention provides a method for evaluating the interactive cognitive ability of a driverless vehicle. The method comprises the following steps: constructing a hierarchical interactive scene library and collecting high-frequency data; constructing a cognitive entropy model; carrying out interactive cognitive calculation based on entropy reduction; and constructing a comprehensive scoring model, and carrying out comprehensive scoring and grading. According to the method for evaluating the interactive cognitive ability of the driverless vehicle, a clear optimization direction is provided for algorithm development, and an innovative implementation path is provided for evaluating the core ability of the driverless vehicle.

The invention relates to the technical field of big data and data visualization, and discloses a cognitive decision-making system fusing BI and AI capabilities, and the system comprises a data access and management module which accesses multi-modal data, carries out the preprocessing of the multi-modal data, and further carries out the full-life-cycle management; the index interaction and analysis module is used for performing index caliber unification processing on the data to obtain standardized index data; a data interaction panel is constructed, and an analysis result is visually displayed; the decision deduction module is used for performing cognitive calculation and dynamic deduction on the standardized index data and generating and outputting decision suggestions; executing a decision behavior to obtain actual effect data, and feeding back the actual effect data to a cognitive calculation process to realize continuous optimization of the module; and the authority workbench module is used for configuring a personalized Web interface according to role authority and integrating multiple functions so as to realize differentiated data access and operation.

The invention discloses a visual interface adaptive adjustment system based on user behavior analysis, and relates to the technical field of behavior analysis. The user behavior data acquisition module is used for acquiring multi-dimensional interaction data and encrypting and transmitting the multi-dimensional interaction data; the cognitive calculation engine module is used for mining behavior patterns and intentions based on deep learning; the environment perception and situation modeling module is used for perceiving an environment and a user state in real time; the self-adaptive interface generation module is used for dynamically adjusting layout, style and interaction; and the intelligent decision-making and continuous optimization module realizes intelligent decision-making through reinforcement learning. According to the system, user behaviors are accurately collected and analyzed, environment changes are sensed, and the visual interface layout and interaction are dynamically optimized; the task time can be remarkably shortened, the error rate is reduced, the user satisfaction degree, the use duration and the service conversion rate are improved, the user stickiness is enhanced, personalized experience is provided for users, and benefits are created for enterprises.

The invention relates to the technical field of the Internet of Things, and discloses an energy efficiency collaborative optimization method based on the Internet of Things, and the method comprises the steps: constructing a value-structure-energy efficiency ternary coupling model, and carrying out the unified modeling of a data value, a system structure and energy distribution; constructing a double-layer recursive optimization architecture which comprises two mutually recursive optimization layers, namely a sensing layer and a relation layer; a self-adaptive tensor decomposition means is introduced to process value-relation-time three-dimensional data; a multi-time scale energy efficiency equalizer is realized, and short-term energy efficiency optimization and long-term energy efficiency planning are processed at the same time; and constructing a hybrid optimization objective function based on information entropy and graph centrality. According to the method, cognitive calculation and graph calculation technologies are fused, co-evolution of data values and system structures is achieved, the data acquisition value is increased by 35% under the same energy consumption condition, or the energy consumption is reduced by 42% under the same data value requirement; the adaptive speed of the system to the environment change is improved by 3.7 times; and the energy efficiency stability is improved by 78%.

The invention provides a multi-mode-driven super-large-scale neural network controller and a danger early warning method thereof. The multi-mode-driven super-large-scale neural network controller comprises an audio and video acquisition module, a multi-sensor module, a data fusion module, an industrial data driving basic model, a controller decision module and an early warning and communication module. The intelligent controller and the method thereof have the beneficial effects that the intelligent controller and the method thereof integrate an audio and video system and multi-sensor information, perform cognitive calculation by using an audio and video neural network compatible with multiple data modalities, and complete environment intelligent evaluation and danger early warning; according to the invention, the audio and video neural network compatible with multiple data modals is applied to multi-modal data cognition and analysis of high-risk environments such as a coal mine for the first time; efficient fusion and cooperative processing of audio and video data and multi-sensor data are realized; an accurate and real-time danger early warning and accident response mechanism is provided, and the safety is remarkably improved; and an embedded platform is used to complete early warning of the intelligent controller.

The invention relates to the cross technical field of computer vision and natural language processing, and discloses a minority image description generation method based on a memory mechanism and cognitive calculation, which comprises the following steps of: extracting visual features through ConvNeXt, constructing national culture knowledge, executing a multi-level cognitive strategy on an input image, and carrying out national feature weighted fusion, so as to obtain the minority image description generation method based on the memory mechanism and the cognitive calculation. The method comprises the following steps of: obtaining a cognitive reasoning result, combining national culture knowledge with the cognitive reasoning result, constructing a logic cue word, splicing the logic cue word, visual features and the cognitive reasoning result, and inputting the spliced logic cue word, visual features and cognitive reasoning result into a multi-modal large-scale language model GLM-4V based on a Transform architecture to generate a description text. According to the method, high-precision understanding and description generation of cultural symbols, clothing utensils and scene semantics in minority images are achieved by fusing a dynamic generation strategy of a multi-modal large language model, and finally a natural language description text with accuracy, richness and culture depth is output.

The cognitive modeling method, device, and application for episodic memory task for early screening of cognitive impairment can effectively activate familiar and recall memory processes related to early damage in the disease. Its neural basis involves brain regions that are susceptible to early Alzheimer's disease. It has the characteristics of automation, strong objectivity, and high interpretability. The design complexity is moderate and it supports fully automatic presentation and scoring on a computer. It is very suitable for large-scale deployment in communities, primary healthcare institutions, or remote environments. The method includes: (1) data collection, which includes two stages: encoding stage and extraction stage; (2) constructing a cognitive computing model, which divides the cognitive process into a series of discrete processing steps. These steps form a decision tree leading to different possible classification results; (3) constructing five logistic regression classifiers and using 10-fold cross-validation to distinguish between cognitively normal individuals and individuals with mild cognitive impairment (MCI). Each model is developed using a different feature set.

A firing neural network computing system and method for brain-like intelligence and cognitive computing. The system includes a model description module, a parameter database, a configuration description module, a configuration manager, a rule manager, a data manager, a network builder, a network manager, an operation manager, a scheduler, a log manager, an operation monitoring module and a graphical display module. The system provides the function of automatically executing synapse and neuron pruning and genesis according to certain conditions and rules, and provides a variety of flexible trigger conditions for starting up pruning and genesis processes as well as rules for executing the processes, which eliminates the burden of neural network developers needing to write synapse and neuron pruning and genesis programs by themselves, thereby effectively solving several problems in existing brain-like firing neural network computing frameworks.