1029 results about "Learning architecture" patented technology

The invention provides a chip production logistics optimization scheduling method and system based on deep reinforcement learning, and relates to the technical field of deep reinforcement learning. An improved bidirectional long-short-term memory network and a graph neural network are used for extracting a material flow time sequence rule, equipment operation cycle characteristics and logistics network topology characteristics, a fusion state representation is generated through a contrast learning network, a layered and collaborative deep reinforcement learning architecture is constructed, an improved encoder-decoder structure is used for generating a strategy vector, and the strategy vector is constructed. And performing hierarchical joint training through a multi-agent execution network, inputting a real-time state and a strategy vector into the trained deep reinforcement learning architecture, generating a scheduling instruction, and performing hierarchical model prediction control in combination with a Bayesian neural network ensemble model and a recurrent neural network prediction model to realize optimized scheduling of chip production logistics.

A modeling system trains a recurrent machine-learned model by determining a latent distribution and a prior distribution for a latent state. The parameters of the model are trained based on a divergence loss that penalizes significant deviations between the latent distribution the prior distribution. The latent distribution for a current observation is a distribution for the latent state given a value of the current observation and the latent state for the previous observation. The prior distribution for a current observation is a distribution for the latent state given the latent state for the previous observation independent of the value of the current observation, and represents a belief about the latent state before input evidence is taken into account.

The invention discloses a multi-modal medical image data intelligent processing system, relates to the field of medical image analysis, and is applied to multi-modal medical image whole-process analysis of CT, MRI, PET, ultrasound and the like. According to the system, different modal image features are extracted and fused through a cross-modal manifold fusion network; a semantic guidance dynamic registration engine optimizes registration parameters to ensure that the registration error is less than or equal to 1.5 mm; the multi-task collaborative diagnosis network realizes multiple tasks such as disease classification; the clinical knowledge embedding and interpretable module generates a structured report and is in butt joint with an HIS system. Meanwhile, the model is optimized through a federated learning architecture, the adaptability of newly added data is improved by more than or equal to 20%, and intelligent processing and analysis of multi-modal medical images are realized.

The present disclosure describes a method comprising receiving a plurality of training images; executing a feature extraction machine learning model to generate a plurality of training embeddings for a plurality of training images each in an embedding space; training a content scoring machine learning model using the plurality of training embeddings to generate performance scores for content items based on embeddings in the embedding space; receiving a set of text; executing the feature extraction machine learning model using the set of text to generate a text embedding in the same embedding space as the training embeddings for the plurality of training images; generating, using the content scoring machine learning model, a text performance score for the set of text using the text embedding in the embedding space; and generating a record identifying the text performance score for the set of text.

The invention provides a cross-platform dynamic security baseline and vulnerability closed-loop repair method and system based on federated learning, and the method comprises the steps: constructing a heterogeneous policy mapping rule base of a Windows registry, Linux sysctl and iOS plist by configuring a semantic analysis algorithm, and forming a multi-operating system policy consistency guarantee mechanism by combining a configuration conflict detection model of formalized verification; a dynamic reinforcement decision engine is designed based on a vulnerability influence surface analysis model driven by a knowledge graph and a double-circulation reinforcement learning architecture (outer-layer strategy exploration and inner-layer parameter optimization); a special evaluation system is constructed for key fields such as electric power and finance, a domestic cryptographic algorithm is deeply fused, and an electric power monitoring system PCSR / IIR / SCAR multi-dimensional index and SM2 / SM4 / SM9 full-stack security scheme is developed. According to the intelligent security baseline reinforcement method, the problems of wide cross-platform strategy gap, extensive vulnerability repair decision and insufficient industry adaptability of a traditional scheme are solved, and the intelligent security baseline reinforcement method supporting dynamic confrontation, accurate adaptation and service fusion is provided.

The present disclosure relates to systems and methods for automatic generation of summaries for record objects using one or more large language models. The system can identify a plurality of electronic activities matched to a record object of a customer relationship management system. The system can generate one or more text strings from the plurality of electronic activities using the large language models. The system can generate a first set of topics by inputting a first set of text strings into the large language models. The system can generate a second set of topics by inputting a first subset of text strings of the first set of text strings into the large language models. The system can transmit one or more topics to a computing device for presentation.

The invention discloses an intelligent psychological intervention system based on multi-modal fusion, which is characterized in that a three-dimensional evaluation system is constructed by integrating speech sentiment analysis, keyboard dynamics monitoring and physiological signal acquisition, and time sequence alignment and feature weighted fusion of multi-source data are realized by adopting a cross-modal Transform model. The core of the system comprises an adaptive intervention engine which defines a multi-dimensional state space based on a hierarchical reinforcement learning architecture, optimizes an intervention strategy through a PPO algorithm, and realizes dynamic emotion interaction in AR and VR scenes in combination with a digital twin training module; according to the clinical decision support system, physiological behavior characteristics and psychological assessment trends are integrated by using a multi-time scale risk prediction model, and a personalized early warning threshold system is constructed, so that the psychological state recognition accuracy is improved, the intervention intensity self-adaptive adjustment response time is shortened, and the high-risk signal early warning timeliness reaches the minute level; and the problems of evaluation hysteresis and strategy stiffness of traditional psychological intervention are obviously improved.

A method includes receiving a pointer to a plurality of images of a web page, the web page configured to display the plurality of images in a first arrangement; retrieving, using the pointer, the plurality of images of the web page; executing at least one machine learning model using the plurality of images from the web page as input to generate at least one image performance score for each image of the plurality of images, the at least one machine learning model trained based on a training set of images labeled based at least on interaction data corresponding to images of the training set of images; and rearranging the plurality of images on the web page to a second arrangement according to the at least one image performance score generated for each image of the plurality of images of the web page.

A machine-learned architecture may predict multiple paths that an object could take in the future without regard to time at which the object may occupy positions identified by one of those paths. These time-invariant paths may be used by an autonomous vehicle to filter detected objects by relevance to an autonomous vehicle's plans, improve prediction of an object's reaction to a vehicle candidate trajectory, determine right-of-way between object(s) and the autonomous vehicle, match detected objects to lanes, and / or improve prediction of odd or out-of-turn object behavior of an object.

A method includes displaying a first user interface including a target image and a task selection element, in response to a selection of the task selection element, displaying a second user interface including a target audience selection element, in response to a selection of a target audience using the target audience selection element, displaying a third user interface including a benchmark selection element, in response to a selection of a benchmark using the benchmark selection element, executing a target audience machine learning model using as input the target image to generate a performance score, calculating a benchmark score for the target image based on the performance score of the target image and performance scores for the images of the benchmark, and displaying a fourth user interface including the target image and the benchmark score.

A system and method for a federated deep learning platform utilizing homomorphically-compressed and encrypted data. The system comprises multiple client devices, each with a local dataset, and a central server hosting a deep learning core. Client devices convert local data into codewords, which are also homomorphically encrypted. The central server processes these encrypted codewords without decryption, preserving data privacy. The platform supports at least two architectural variants: a conventional Transformer trained on codewords, and a Latent Transformer operating on latent space vectors. Both variants eliminate the need for embedding and positional encoding layers. The system aggregates encrypted model updates from clients, enabling collaborative learning while maintaining data confidentiality. Additional features comprise differential privacy implementation and adaptive federated optimization techniques. This innovative approach allows for efficient, privacy-preserving distributed learning across diverse datasets, addressing key challenges in federated learning such as data heterogeneity, non-IID distributions, and communication efficiency.

Presented herein are systems and methods for the employment of machine learning models for image processing as may be performed by computing devices associated with an end user. A method may include obtaining video data comprising a plurality of frames including a document of a document type. The method may include executing an object recognition engine of a machine-learning architecture using image data of the plurality of frames, the object recognition engine trained to detect edges of documents. The method may include identifying, based on the edge detection, a plurality of boundaries for the document. The method may include validating, based on the plurality of boundaries, the document as the document type. The method may include transmitting via one or more networks, to a computer remote from the computing device, responsive to the validation of the type of document, the image data for the plurality of frames depicting the document.

The invention relates to the technical field of shopping guide, in particular to a multi-mode user intention understanding and personalized shopping guide generation method and system, and the method integrates vision, voice and touch screen data to construct a commodity and user intention knowledge graph. Through a cross-modal contrast learning architecture and deep feature learning of user intentions, precise commodity matching and sorting are realized; and in combination with a knowledge-enhanced semantic alignment model, generating a personalized recommendation verbal skill, and making a shopping guide strategy by using a sequence-to-sequence neural network. Complementary information among different modal data can be captured, a unified semantic representation space can be learned, and the accuracy of user intention understanding is greatly improved; the system breaks through the technical barriers, significantly improves the intelligent shopping guide capability of the e-commerce platform, rebuilds the shopping experience of the user, shows huge commercial value, and injects new power for industry development.

The present disclosure relates to systems and methods for automatic generation of datasets for record objects using one or more large language models. A system can identify a plurality of electronic activities. The system can generate, using one or more large language models, a first set of text strings from the electronic activities. The system can store, in one or more data structures, a first association between the first set of text strings and the record object. The system can identify a second electronic activity. The system can generate, using the one or more large language models, a second set of text strings based on data corresponding to the second electronic activity and the first set of text strings. The system can store, in the one or more data structures, a second association between the second set of text strings and the record object.

The invention discloses an expressway lane control method based on real-time traffic flow. The expressway lane control method comprises the following steps of performing fusion sensing through a distributed optical fiber vibration sensor array and a millimeter wave radar; constructing a space-time diagram convolutional network model, and dynamically mapping the road topological structure into a lane connection diagram with variable weight; a layered reinforcement learning architecture is adopted, and an emergency lane dynamic management mode is triggered; and deploying a digital twin sand table to perform multi-resolution simulation verification, and when the actual traffic flow deviates from a predicted value by more than 15%, starting a causal inference engine to backtrack a decision link and update an agent strategy library. According to the method, global dynamic lane optimization is realized through a multi-source perception fusion and hierarchical reinforcement learning architecture, simulation data shows that the morning peak traffic volume is improved by 29.3%, and the accident scene traffic efficiency is improved by 28.6%; a digital twinning and causal inference engine is adopted, the strategy updating response time is shortened to 30 seconds, and the congestion propagation distance is reduced by 35%.

The invention discloses a metal cutting process parameter optimization analysis method based on machine learning, and particularly relates to the field of machine learning. Comprising multi-dimensional process parameter feature extraction and preprocessing, cutting state intelligent identification based on integrated learning, dynamic process parameter sensitivity analysis and weight calculation, process parameter intelligent optimization under a multi-target constraint condition, and adaptive parameter adjustment and real-time control strategy. According to the method, the interaction relationship between complex nonlinear features and process parameters in the cutting process is comprehensively captured, and accurate and intelligent recognition of different cutting states such as normal cutting, tool abrasion and abnormal flutter is achieved through a three-layer integrated learning architecture; the technical bottlenecks that an existing system lacks real-time self-adaptive adjustment capacity and is low in process optimization efficiency are overcome, pertinence and effectiveness of parameter adjustment are ensured, and the technical current situation that machining quality fluctuates and repeatability is poor due to traditional fixed parameters is changed.

The invention provides an intelligent water conservancy digital twinborn simulation system based on multi-source data, and belongs to the technical field of digital monitoring. Minute-level data acquisition and transmission are realized by constructing a space-air-ground three-dimensional sensing network and combining edge intelligent preprocessing, data acquisition, noise reduction and abnormity identification are realized by utilizing sensors such as millimeter wave radar and laser radar, and the system is used for realizing data acquisition and transmission. Dynamic data fusion and intelligent calibration are carried out, and second-level alignment and credible verification of data are realized by means of a space-time calibration algorithm, Kalman filtering and a block chain evidence storage technology; a hybrid simulation and intelligent decision model is established, a physical model, a machine learning architecture and a dynamic threshold decision tree are adopted, flood routing minute-level prediction and emergency response are realized, and the system improves water conservancy monitoring prediction precision and emergency response efficiency.

The invention relates to the technical field of industrial production lines, and discloses an industrial production line multi-device dynamic collaborative scheduling method based on reinforcement learning, comprising the following steps: S1, modeling a three-dimensional state space; s2, hierarchical reinforcement learning architecture; and S3, edge-cloud cooperative execution. According to the industrial production line multi-device dynamic collaborative scheduling method and system based on reinforcement learning, device states, task constraints and resource occupation are integrated into a structured matrix through three-dimensional state space modeling, and a global decision-making layer captures production time sequence dependence by using a bidirectional long-short-term memory network; modeling equipment space association and process constraints through a graph attention network, and generating a global strategy including task allocation, capacity adjustment and resource pre-allocation; and after the edge layer detects the dynamic event, the cloud platform generates a candidate scheme through Monte Carlo tree search, and realizes dynamic event response and multi-target collaborative optimization by combining multiple targets such as global value network evaluation task completion time and equipment load balancing.

The codebook-based homomorphic compression system is a novel approach that combines data compression and homomorphic encryption to enable efficient and secure computation on compressed data. It involves quantizing the input data, generating an optimized codebook using techniques like Huffman coding or deep learning, and compressing the data by replacing each value with its corresponding codeword. The compressed data is then encrypted using a homomorphic encryption scheme, allowing computations to be performed directly on the encrypted compressed data without decryption. The system includes a deep learning architecture trained on and operating on the homomorphically compressed and encrypted data, enabling inference and decision-making while maintaining data privacy. This approach combines the benefits of data compression, homomorphic encryption, and deep learning, enabling efficient storage, transmission, and secure computation on compressed data streams.

The invention relates to the field of particle size distribution measurement, in particular to a particle size distribution measurement method and system based on machine learning and a light scattering method.The method comprises the steps that a to-be-measured particle sample is irradiated through a light source, and meanwhile light scattering intensity data of different scattering angles and different wavelengths are collected through a multi-angle scattered light detector; constructing a multi-modal deep learning model, and forming an integrated learning architecture in combination with a support vector machine model and a random forest model; inputting the preprocessed multi-wavelength light scattering intensity data into the trained integrated learning architecture to obtain a preliminary particle size distribution result of the to-be-detected particle sample; based on the improved Mie scattering theory, a high-precision theoretical model considering the particle shape, the refractive index, the surface roughness and the internal structure is constructed; an improved Bayesian optimization algorithm is adopted, and the initial particle size distribution result is optimized and adjusted by minimizing the error between the theoretical light scattering intensity and the actually collected light scattering intensity, so that a corrected particle size distribution result is obtained.

The invention provides an investment decision-making system and method based on federated learning and multi-modal risk control, and the system comprises a data collection layer which is used for integrating satellite remote sensing data, supply chain Internet of Things data and financial market data; the federal processing layer is used for constructing a distributed training framework based on a FedAvg algorithm and realizing cross-mechanism data fusion and global model parameter updating through homomorphic encryption and differential privacy; the intelligent decision-making layer adopts a TD3 reinforcement learning engine and adopts a dynamic gating mechanism to adjust fusion weights of texts, time sequences and image modalities in real time; and the execution monitoring layer is used for detecting transaction execution deviation in real time through a digital twinning technology, realizing abnormal response by adopting a CUSUM algorithm, and dynamically adjusting a federal learning rate through Bayesian optimization. The key technical defects of data islands, tail risk response delay, insufficient cross-market linkage analysis and the like of a traditional investment system are overcome through an innovatively designed federated learning architecture and a time-varying risk control model.

The invention belongs to the technical field of Internet of Vehicles, and particularly provides a federated learning optimization system and method for an intelligent network connection automobile scene, which improves an asynchronous federated learning framework, divides a plurality of vehicles into different clusters due to the isomerism of computing power, and adopts a cloud-edge-end three-layer federated learning framework, so as to realize the optimization of the federated learning of the vehicles. An edge server is deployed on a roadside unit RSU, received model parameters are subjected to pseudo-asynchronous aggregation in a self-adaptive dynamic time window, the received model parameters are subjected to real-time aggregation at a cloud end, the system model training efficiency and quality are improved, meanwhile, the data security is ensured, the user privacy is protected, and the user experience is improved. And the efficiency of the federated learning optimization system and the quality of the global model in the intelligent network connection automobile scene are ensured.

The present disclosure relates to systems and methods for automatic generation of summaries for record objects using one or more large language models. The system can identify a plurality of electronic activities matched to a record object of a customer relationship management system. The system can generate one or more text strings from the plurality of electronic activities using the large language models. The system can generate a first set of topics by inputting a first set of text strings into the large language models. The system can generate a second set of topics by inputting a first subset of text strings of the first set of text strings into the large language models. The system can transmit one or more topics to a computing device for presentation.

Deep learning tools such as convolutional neural networks (CNNs) and transformers have spurred great advancements in computational biology. However, existing methods are constrained architecturally in context length, computational complexity, and model size. This application introduces a sub-quadratic architecture for modeling, which combines projected gated convolutions and structured state spaces to achieve local and global context with, for example, single-nucleotide resolution. These models outperform CNN-, GPT-, BERT-, and long convolution-based models in many tested genomics tasks without pre-training and with 4×-781× fewer parameters. In the proteomics domain, these models similarly outperform pretrained attention-based models, including ESM-1B and TAPE-BERT, on remote homology prediction without pre-training and while using 3,308×-23,636× fewer parameters.

The invention relates to the technical field of medical image processing, in particular to a medical image enhancement processing method and system based on deep learning, and the method comprises the steps: selecting an image enhancement model according to whether the number of samples needs to be expanded or not; if the number of samples needs to be expanded, performing image enhancement by adopting a generative adversarial network, otherwise, performing image enhancement by adopting an auto-encoder; a federal learning architecture is adopted to train a model, and the model is trained on a plurality of scattered devices or servers. The generative adversarial network is used to generate diversified medical image samples, and the diversity of the samples is improved. And the auto-encoder extracts key features and generates new samples, so that the diversity of data is increased. Through a federal learning architecture, by using dispersed data sources, the diversity of training samples of the model is enhanced, and meanwhile, data privacy is protected. And the diversity of samples is effectively increased, so that the generalization ability of the model is improved, and the method has higher robustness and accuracy in practical application.

A real-time anomaly detection system for high-volume managed file transfers (MFT), consisting of: a secure, hardware-accelerated monitoring unit configured to interface with a managed file transfer server and intercept file transfer session data at wire-speed; a metadata extraction engine embedded in the hardware-accelerated unit, the metadata extraction engine configured to analyze protocol-specific session attributes, including, but not limited to, file size, transfer duration, encryption status, source and destination endpoints, transfer frequency, and payload entropy; a contextual AI inference engine communicatively coupled to the metadata extraction engine, the AI inference engine comprising a deep learning model trained on labeled historical MFT activity logs to detect contextual deviations from normative behavior; a federated learning architecture with a plurality of edge nodes, each hosting a local anomaly detection model trained on localized transmission metadata and configured to synchronize with a central aggregator using differentially private gradient updates; an Explainable AI (XAI) subsystem integrated into and configured to generate human-readable anomaly justifications, feature importance maps, and threat categorization labels; and a policy orchestration module configured to dynamically execute pre-configured or AI-based security responses, where the security responses include selective session termination, quarantining of transferred files, generation of alerts, or redirection of MFT workflows.

The invention relates to the field of medical image processing, in particular to a lung cancer PET-CT fusion segmentation method and system based on multi-modal feature comparative learning, and the method comprises the steps: firstly extracting PET and CT image features, projecting the features to a shared semantic space through a semantic guide type symmetric comparative learning architecture, obtaining key region features through a focus adaptive attention sampling mechanism, and carrying out the segmentation of a target region; optimizing feature representation through a cross-modal feature difference self-calibration mechanism, constructing a multi-scale feature pyramid, fusing features of different scales by using a multi-scale hierarchical contrast learning mechanism, and performing self-supervised learning by combining an anatomical guidance self-supervised contrast learning enhancement module and using a CT anatomical structure, so as to further reinforce the features; a high-precision lung cancer lesion segmentation result is generated through a decoder network, the Dice coefficient is increased from 0.78 to 0.91, and the detection rate of lesions below 10 mm is increased from 65% to 87%. A novel efficient and accurate image processing method is provided for lung cancer diagnosis.

A machine-learned architecture may predict a set of spatially-diverse paths that an object may take in the future. The paths generated by this architecture may be time-invariant (e.g., not identifying a time at which the object may occupy a position along one of these paths) but can be used by a second machine-learned model to predict progress in time along these paths. This segregation of the spatial paths and progress in time along the paths improves the accuracy of the ultimate prediction and better captures rare object behavior.

The present disclosure is directed toward systems, methods, and non-transitory computer readable media that provide a contextual query answering system that trains and implements a unique machine learning architecture to generate accurate domain-specific contextual responses. For example, the disclosed systems receive a contextual query indicating a software context of a computer application within a software-specific domain. The disclosed systems utilize a context retrieval model to generate query embeddings from the contextual query and data segment embeddings from data segments of stored digital documents. Further, the context retrieval model determines relevant digital documents from among the stored digital documents based on comparing the query embeddings and the data segment embeddings. The disclosed systems provide the relevant digital documents to a response generator model to generate a contextual response within the software-specific domain.

The invention relates to the technical field of injection molding process control, and discloses an injection molding parameter self-learning collaborative regulation and control system which is characterized in that multi-source parameters such as temperature and pressure are acquired in real time through a data acquisition module, and the multi-source parameters are mapped to a low-dimensional manifold space through a dynamic manifold construction module to decouple a nonlinear coupling relationship; the chaos synchronization control module suppresses external disturbance based on master-slave system design, and the distributed optimization module realizes multi-target real-time decision by adopting a split Bregman algorithm; the execution module converts the optimization instruction into a control signal and triggers closed-loop feedback; and the federated learning architecture realizes cross-device strategy migration by extracting continuous homology features of a manifold space and combining a Hausdorff distance and homeomorphic mapping, and drives a global model to update. According to the invention, the technical problems of difficult dynamic regulation and control of nonlinear parameters, insufficient real-time performance of multi-constraint optimization, low cross-scene repeated parameter adjustment efficiency and lack of self-adaptability of long-term process drift in the injection molding process are solved.