61 results about "Heuristics" patented technology

The present invention discloses a modular AI-based system for knowledge instantiation and evidence synthesis, unifying five modules: Knowledge Instantiation, ingesting anchor knowledge via retrieval- augmented generation and neuromorphic prompting; Case-Based Agent Generation, orchestrating domain constraints, performance criteria, and sub-task decomposition; Secondary Epistemogenesis, spawning sub-agents with inherited heuristics; Workflow Orchestration and Evidence Synthesis, merging outputs, tracking metadata, and storing final artifacts; and Reinforcement Learning from Human Refinement, capturing feedback and edits. The method for output generation initiates by defining knowledge requirements, indexing public and proprietary knowledge, validating output structure, drafting content, and refining outputs via sub-agent spawning and human oversight. The system ensures domain alignment, concurrency control, and persistent versioning. By combining specialized knowledge sources, multi-agent concurrency, and iterative human feedback loops, it dynamically adapts to evolving knowledge requirements. This invention emphasizes reliability, traceability, and AI-driven outputs in regulated industries, delivering trustworthy outcomes anchored in expert knowledge.

The invention relates to the field of automatic algorithm design, in particular to an automatic algorithm design method for solving a vehicle path planning problem by utilizing a large language model, which comprises the following steps of: S1, constructing an AutoDH framework, and defining a decision form of an intelligent agent; s2, defining heuristics in an LLM pool, an improvement pool and a disturbance pool; s3, acquiring state information of the CVRPs, and intelligently selecting a heuristic mode through two-stage MDP; s4, designing a reward mechanism fusing solution quality improvement, heuristic time cost and LLMs API calling cost; and S5, training the intelligent agent to optimize the solution of the CVRPs. The method can intelligently select the most cost-effective heuristic algorithm according to the state information of the current CVRPs, and improves the flexibility and adaptability of algorithm design.

Systems and methods for training and using large language models to respond to queries and cascading to new models when their performance falls below a threshold during the response generation cycle are described. The methods involve receiving and analyzing a query using heuristics to determine the query's categories and a level of granularity at which its response is to be evaluated. Selecting a language model based on the query analysis and during the generation of the response, evaluating the model's performance for its ability to predict the next segment in a response. The methods score the evaluation and use the score to determine whether the language model's performance exceeds a confidence threshold. If it does not, then during the response is being generated, e.g., at a point before the response generation is completed, cascading from the model to a different model that can perform at a higher confidence level.

Disclosed apparatuses and methods for trajectory and collision prediction comprising a processor operable to generate, using a machine-learning (ML) model, a ML predicted trajectory of a vehicle and a ML model confidence of the ML predicted trajectory, generate, using a heuristics model, a heuristic predicted trajectory of the vehicle, determine whether the ML model confidence satisfies a metric, in response to determining that the ML model confidence satisfies the metric, apply the ML predicted trajectory as a predicted trajectory, and in response to determining that the ML learning model confidence fails the metric, apply the heuristic predicted trajectory as the predicted trajectory.

A data platform monitors a compute environment by performing multi-stage heuristic analysis of event data representing a plurality of events occurring within the environment. The platform utilizes multiple event analyzers, each configured according to a distinct analysis heuristic, to evaluate different subsets of the event data and generate corresponding output signals. A higher-level event analyzer applies a further heuristic to the multiple output signals to generate a composite alert signal, indicating whether the combination of analyzed events collectively represents a security intrusion or other anomalous condition of sufficient severity to warrant alerting. Based on the composite alert signal, the platform performs an alert-based operation, such as generating a user-facing alert, initiating an automated mitigation, or updating a contextual model of system behavior. By combining the analytical outputs of heterogeneous heuristics, the disclosed architecture enhances the accuracy and contextual relevance of automated intrusion detection within complex computing environments.

This disclosure describes techniques and architecture provide automated contribution analysis for “why question” style NLQ answering, e.g., “why is revenue down in North America Q1 2022.” In particular, the techniques described herein combine multiple signals together including, for example, frequency of use of combinations of dimensions in previous NLQs (warm-start), statistical information about columns (e.g., entropy), correlation / co-occurrence between pairs of dimension columns, and correlation between dimensions and dates. This information is used with a set of heuristics and rules to pick the best set of dimensions as contributing factors for a particular metric over a particular time period and present an automatic contribution analysis to the users to give them insights into their data.

There is provided a system and methods to determine an improved solution to a Mixed Integer Problem (MIP) using a quantum-assisted MIP solver. The methods are performed by a digital processor in communication with a quantum processor. Methods include: selecting at least one feasible solution determined by an MIP solver, determining a first sub-problem of the MIP based on the at least one feasible solution; casting the first sub-problem as Binary Quadratic Models (BQMs); solving the BQMs using the quantum processor to generate sample solutions; determining a second sub-problem based on at least the sample solutions, and obtaining a current solution to the MIP by evaluating the second sub-problem; and updating an incumbent solution if the current solution improves over the current incumbent solution. The quantum-assisted MIP solver uses hybrid crossover and mutation heuristics to improve the convergence time and accuracy of solutions obtained using Branch-and-Cut solvers.

A method for operating a quantum error correction (QEC) code on a quantum computing system (QCS) is disclosed. The QCS includes a set of qubits and a set of couplers. An indication of a set of dropouts is received. Each dropout corresponds to a qubit that is non-functional or a coupler that is non-functional. The dropouts define a set of non-functional qubits, a set of functional qubits, a set of non-functional couplers, and a set of functional couplers. The QEC code is generated based on a set of heuristics, the set of functional qubits, and the set of functional couplers. The QEC code operates on the functional qubits. Each functional coupler provides a coupling between a pair of functional qubits. A quantum algorithm is executed that includes employing the QEC code to protect a set of logical qubits formed by the first subset of functional qubits from logical errors.

Systems, computer program products, and methods are described herein for synthetic intrusion data generation and remediation via machine learning. The present disclosure includes training a first machine learning model using a plurality of malicious code segments from a code repository, generating, using the first machine learning model, a predetermined number of generated code segments, training a second machine learning model using the plurality of malicious code segments, generating, by using the second machine learning model, at least one generated heuristic mitigation resource for the generated code segments, analyzing, via a static heuristics analysis, stored code on an endpoint device, and applying the at least one generated heuristic mitigation resource upon a first condition wherein the static heuristics analysis identifies a malicious activity.

The invention relates to the technical field of new energy power prediction, and discloses a new energy power prediction method and system. According to the method, through multi-source data preprocessing and correlation screening, key influence factors are focused to construct an initial prediction model; a dynamic mixed meta-heuristic optimization strategy is adopted, global efficient exploration is carried out on the upper layer model, local fine optimization is carried out on the lower layer model, collaborative optimization is carried out on hyper-parameters of the initial prediction model in stages, and finally a target prediction model is formed and used for real-time power prediction. According to the method, through the synergistic effect of the upper-layer model and the lower-layer model of dynamic mixed meta-heuristic optimization, the advantages of different algorithms are combined, and it is ensured that the hyper-parameter optimization process is more likely to converge to a globally optimal solution. The technical problems that an existing single optimization algorithm often has limitation and is prone to falling into local optimum or low in convergence speed are solved.

A process, system and medium for detecting anomalous authentication requests to a protected resource during an authentication journey, in order to regulate step-up authentication are described. The process includes obtaining request features from the authentication request that triggered the authentication journey. The process includes processing, by an ensemble of Machine Learning (ML) models and a set of rule-based heuristics, a set of features based on the request features, the set of features associated with a userID. The process includes deriving risk sub-scores for each ML model and each heuristic. The process includes deriving a risk score based on the risk sub-scores. The process includes determining that the risk score exceeds an explanation-triggering threshold. The process includes providing, to a node in the authentication journey, the risk score with an explanation of the risk score. The system and medium are configured to execute the process, and configured to provide the explanation.

A method, computerized apparatus and computer program product, the method comprising: obtaining a collection of entities including a source code file and a non-source-code file; identifying a bundle within the collection of entities; generating a call and dependency graph of the collection of entities, the graph comprising two or more nodes each associated with an entity, and at least one edge indicating a calling relationship between the nodes, said generating comprising: generating an initial call and dependency graph, indicating reachability of each of the collection of entities starting from an entry point; and pruning the initial call and dependency graph by reducing an edge according to collected heuristics, to obtain a pruned call and dependency graph; obtaining information about vulnerabilities associated with the bundle or any of the collection of entities; and providing an indication of a reachable vulnerability and an unreachable vulnerability in the pruned call and dependency graph.

A system method for performing distributed training of models. The method includes co-optimizing memory optimizations with parallelism to increase model training throughput under a memory constraint by orchestrating a plurality of optimizations to utilize system resources in consideration of computation, communication and memory footprint. The system can include an overlap-centric schedule template that determines granularity and order of how techniques utilized by the plurality of optimizations are applied to a model. The overlap-centric schedule template mitigates tuning complexity by applying heuristics to orchestrate optimizations in an overlapped manner.

The invention discloses a two-stage reinforcement learning SAT solving method for structure perception and a related device, and belongs to the field of chip design. Chip signal polarity and logic consistency information is introduced in a variable initialization stage, an initial feature vector for structure perception is constructed, and the limitation that traditional graph modeling can only represent a local logic relation is broken through; a two-stage reinforcement learning framework composed of a main path and a sampling path is designed, the sampling path is used for exploring a chip logic solution space, the main path is used for stably utilizing an optimal verification strategy, and the problems of unstable training convergence, single strategy and the like are effectively relieved. Based on a reward-guided feature updating mechanism, dynamic evolution of chip signal features and enhanced feedback are explicitly bound, the interpretability of a solver is improved, and the difference between neural reasoning and chip logic symbol reasoning is reduced. A signal assignment decision is executed based on GNN multi-round message passing results, and a verification track is continuously optimized, so that efficient chip design verification capability can be obtained without depending on an artificial heuristic mode.

The present application relates to the technical field of GPU computing optimization, and in particular to a sparse attention computing method, device and medium for GPU, wherein the method realizes the high efficiency of long context reasoning through the geometric perception sparse attention framework of ball hashing, and combines a large-scale parallel hashing optimization algorithm and a load adaptive computing kernel. Compared with the existing sparse attention methods based on heuristics or gradient learning, the present application realizes higher retrieval recall rate, lower preprocessing overhead and efficient hardware adaptation to irregular sparse patterns.

Aspects of the disclosure relate to arranging a pick up and drop off locations between a driverless vehicle and a passenger. As an example, a method of doing so may include receiving a request for a vehicle from a client computing device, wherein the request identifies a first location. Pre-stored map information and the first location are used to identify a recommended point according to a set of heuristics. Each heuristic of the set of heuristics has a ranking such that the recommended point corresponds to a location that satisfies at least one of the heuristics having a first rank and such that no other location satisfies any other heuristic of the set of heuristics having a higher rank than the first rank. The pre-stored map information identifying a plurality of pre-determined locations for the vehicle to stop, and the recommended point is one of the plurality of pre-determined locations. The recommended point is then provided the client computing device for display on a display of the client computing device with a map.

The application relates to the technical field of binary program analysis and rewriting, and discloses an immediate number symbolization judgment method and system of a binary rewriting program, which comprises a preprocessing step, a basic evidence analysis step, an intermediate evidence analysis step, a high-level evidence analysis step and a fusion decision step. The preprocessing layer carries out structural analysis and semantic modeling on a binary file, extracts candidate immediate numbers and meta information, the basic evidence layer carries out preliminary weighting based on explicit rules, the intermediate evidence layer carries out quantitative evaluation through heuristics and statistical patterns, the high-level evidence layer carries out high-precision analysis on the immediate numbers with semantic ambiguity, the fusion decision layer summarizes the evidences at all levels, forms a comprehensive tendency value through weighted calculation, classifies the immediate numbers, and generates a final symbolization strategy. The application significantly improves the accuracy and interpretability of symbolization judgment, balances the analysis precision and processing efficiency, and is suitable for binary files of different compilers, platforms and file stripping states.

Systems, methods, and computer program products for identifying zones of interest in text transcripts. An application may receive input specifying a text statement type and determine a plurality of heuristics for identifying statements of the statement type in transcripts. The application may determine, based on a first heuristic, a first text statement of the statement type. The application may generate, based on a clustering algorithm, a plurality of additional statements of the statement type. The application may receive a first text transcript. The application may identify, based on a second heuristic, a first text statement in the first text transcript, where the first text transcript statement is of the statement type. The application may generate a graphical indication that the first text transcript statement is of the statement type, and display the first transcript statement and the graphical indication on a display.

This invention discloses a single-arm combined device scheduling method based on hierarchical attention graph deep reinforcement learning. This method aims to address the problems of existing rule-based methods relying on artificial heuristics, and the complexity of Petri net models while neglecting topological properties. This invention proposes a State Extraction Graph (SEG) to simplify the modeling process for multi-variety wafer manufacturing, and its graph structure is simpler than Petri nets. Based on SEG, a graph deep reinforcement learning framework is developed, integrating a graph neural network based on a hierarchical attention mechanism (called A-GIN) as a feature extractor to better extract the topological properties of the graph. The single-arm robot agent interacts with the environment, autonomously learns and optimizes wafer release and task sequences, achieving end-to-end policy optimization.

Aspects of this technical solution can generate, according to one or more first environment metrics, a three-dimensional (3D) model including a first surface corresponding to one or more physical ways through a physical environment, the one or more first environment metrics indicative of boundaries of the one or more physical ways, generate, according to one or more second environment metrics, one or more geometric two-dimensional (2D) objects on the first surface, the second environment metrics indicative of the one or more physical ways, identify, according to one or more viewpoint metrics indicative of cameras of a physical object configured to move along the one or more physical ways, one or more viewpoints oriented to capture corresponding portions of the 3D model, and render, from the one or more corresponding portions of the 3D model, one or more 2D images each corresponding to respective ones of the viewpoints.

Systems and methods for enterprise production scheduling using a self-learning, resilient Reinforcement Learning (RL) agent. The RL agent interacts with a simulated production environment modeled as a dynamic graph, enabling efficient handling of complex multi-stage scheduling dependencies. Through iterative training, inference, and continuous learning modes, the agent autonomously learns optimal scheduling policies, adapts to evolving production conditions, and incorporates user preferences. The system includes components such as a data profiler for historical analysis, a synthesizer for training data generation, and an initializer for environment setup. The RL agent generates multiple feasible schedules, refines its policy based on feedback, and significantly reduces computational overhead compared to traditional heuristics and genetic algorithms.

Prior methods of state estimation, based on a constrained optimization problem with equality and / or inequality constraints, rely on penalty-based heuristics which can produce very large weight values, resulting in ill-conditioning of the gain matrix. Disclosed embodiments of state estimation convert the constrained optimization problem into an unconstrained convex optimization problem in which violated equality and / or inequality constraints are represented as parameterized potential functions, each comprising a center-of-attraction parameter. This unconstrained convex optimization problem can be iteratively formed, using successively updated values for the center-of-attraction parameters, and solved, until no equality and / or inequality constraints are violated, to produce a final estimated state. This final estimated state may then be used to control the system being monitored, such as a power system.

In an iterative loop a machine learning model (MLM) is trained (TR) with labeled data (LD), thereby forming a trained machine learning model (T-MLM), and then deployed (DP). The trained machine learning model (T-MLM) then receives as input at least pairs of equipment identifiers contained in unlabeled data (UD) and calculates (CL) predictions (P), wherein the predictions (P) contain at least one match prediction (P-M) for a pair of equipment identifiers indicating that the equipment identifiers refer to a same industrial equipment, and wherein the predictions (P) contain in particular at least one different prediction (P-D) for a pair of equipment identifiers indicating that the equipment identifiers refer to different industrial equipment. An inaccuracy detector (ID) using an inaccuracy heuristics (IH) containing in particular probability thresholds for correct predictions, detects (DT) accurate and inaccurate predictions among the predictions (P), and collects the accurate predictions as automatically labeled data (ALD). These are used to boost (BT) existing labeled data (ELD), which is used in a next iteration as the labeled data (LD) for the training operation (TR). In contrast to rule-based approaches, where domain experts manually create a large number of rules, the inaccuracy heuristics (IH) can be primarily derived from data driven analyses which do not require a big investment of time from domain experts.

Methods, systems, and computer-readable storage media for providing a set of heuristics representative of training data that is to be used to process a ML model through a training pipeline, the training pipeline including multiple phases, determining a set of time estimates by providing the set of heuristics as input to a training heuristics model that provides the set of time estimates as output, each time estimate in the set of time estimates indicating an estimated duration of a respective phase of the training pipeline, receiving, during processing of the ML model through the training pipeline, progress data representative of a progress of processing of the ML model, determining a set of status estimates including a status estimate for each phase of the training pipeline based on the progress data, and transmitting the set of time estimates and the set of status estimates for display.

A system method for performing distributed training of models. The method includes co- optimizing memory optimizations with parallelism to increase model training throughput under a memory constraint by orchestrating a plurality of optimizations to utilize system resources in consideration of computation, communication and memory footprint. The system can include an overlap-centric schedule template that determines granularity and order of how techniques utilized by the plurality of optimizations are applied to a model. The overlap-centric schedule template mitigates tuning complexity by applying heuristics to orchestrate optimizations in an overlapped manner.

A training optimization system implements algorithmic solutions to solve the conjoined problem of accelerator architecture search and model partitioning for distributed training. The system makes the multi-dimensional optimization space of architecture search and device placement tractable by reducing the number of accelerator architectures explored through area-based heuristics and employing a novel integer linear program (ILP), the size of which is dependent only on the number of operators. The ILP scheduling optimization also explores the partitioning of operators across cores, known as intra-operator parallelism. Despite the vast space, the ILP described herein requires significantly less time to perform the optimizations across all explored accelerator configurations. Based on the optimal backward and forward pass latencies, the system leverages a novel dynamic programming (DP) approach to determine the device placement and model partitioning scheme.

The approaches presented here involve clustering geometric primitives, such as triangular areas, within a mesh representation, which is advantageous for downstream processing. In at least one embodiment, clustering can be partially based on the spatial positions of the geometric primitives, for example, to minimize the surface area of ​​the bounding boxes around clustered primitives or to minimize the overlap between bounding boxes. Other factors, such as connectivity within or between clusters, can also be considered. A cost function can be used that comprises a weighted combination of cost terms, where these cost terms can be selected, in part, based on the type of downstream processing to be performed or on hardware requirements / optimizations for performing the processing.

The invention relates to the technical field of on-machine measurement, and discloses a heuristic and TS-based OMI path optimization method, system, medium and equipment, and the method comprises the steps: obtaining on-machine measurement points to obtain an initial measurement point set, and constructing an initial flag bit set; combining the two initial sets to construct an initial measurement path, and updating a measurement point set according to the initial measurement path; iteratively optimizing the measurement path and the measurement point set, in each iteration optimization, selecting points from the measurement point set according to the distance between the measurement path and the points in the measurement point set, adding the points into the measurement path, and updating the flag bit set; obtaining a candidate path until the measurement point set is empty; repeatedly generating candidate paths by using the same method until the flag bit set contains all the measurement points, and selecting the shortest path from all the candidate paths as an initial path; and optimizing the initial path by using a tabu search algorithm to obtain a final measurement path. According to the method, the optimal measurement path can be ensured under different measured characteristics, and the solving efficiency is improved.

A system and method of source and type-independent organization, presentation, and management of security detection artifacts, with heuristic and ML-based (machine learning) prioritization algorithms, providing for unified threat detection management and accurate prioritization without requiring an expert user. This is accomplished using an alternative presentation paradigm underpinned by a combination of heuristics and machine learning. The benefits include greater information gain; improved speed of decisioning cybersecurity detection data; elimination of the split-brain problem where different classes of detection artifacts are stored and managed in different systems; and reduced need for expert users. In cases where machine learning and / or anomaly detections are not available for clustering.

Prior methods of state estimation rely on penalty-based heuristics to enforce inequality constraints, which can produce very large weight values, resulting in ill-conditioning of the gain matrix. Disclosed embodiments of state estimation convert the inequality-constrained optimization problem into an unconstrained optimization problem in which violated inequality constraints are represented as parameterized potential functions, each comprising a center-of-attraction parameter. This unconstrained convex optimization problem can be iteratively prepared, using successively updated values for the center-of-attraction parameters, and solved, until no inequality constraints are violated, to produce a final estimated state. This final estimated state may then be used to control the system being monitored, such as a power system.