45 results about "Truth value" patented technology

Provided in the embodiments of the present application are a large-model performance assessment method and a related device. The method comprises: acquiring structured text description of a test material, wherein the structured text description comprises information of the test material in a plurality of dimensions; inputting the structured text description and a test question into a large language model, so as to acquire an evaluation true value; inputting the test material and the test question into a large model under test, so as to acquire an answer under test of the test question; and inputting the evaluation true value, the test question and the answer under test into the large language model, so as to acquire an assessment result for the answer under test.

A computer-implemented method for facilitating reasoning under conditions of uncertainty includes receiving input including a set of logic formulas, a set of intervals representing lower and upper bounds on the truth values of the formulas in the set of logic formulas, and a query formula. The logic formulas can be converted into a logical credal network (LCN) representation and a factor graph representation of the LCN representation can be created. The method can output a probability interval [l, u] such that l≤P(q)≤u, where P(q) represents a query for a given probability interval.

A tailored interactive language learning system that teaches an individualized set of vocabulary words to users through interactive avatars and stories. The interaction is modeled through probabilistic rules in a semantic network and neural network having objects and relations. Dialog and narration is generated dynamically based on the state of the interactive story model using phrasal rewrite rules and neural network implementiung a four-valued logic system in which truth values of the objects and relations are encoded as true, false, defined, and undefined in a single memory array.

Systems, methods, apparatuses, and computer program products for verifying positioning accuracy. A method may include sending, by a test equipment, a request to measure position coordinates to a device, receiving, by the test equipment, a model reasoning including at least one position coordinate from the device, and determining, by the test equipment, whether a difference between the model reasoning and at least one ground truth value is within a threshold for a defined number of samples, wherein the at least one ground truth value comprises at least one known position coordinate of the test point.

The invention discloses a question and answer method, a training method of a question and answer large model, related equipment and a program product, and adopts a retrieval enhancement generation scheme based on the large model, and answer content and knowledge reference sources can be generated at the same time. The adopted large model is trained through reinforcement learning in advance, and in order to improve the correctness and factuality of knowledge reference sources generated by the large model, a reward function in the reinforcement learning process is determined based on reference consistency and / or fact consistency. The reference consistency is used for encouraging the knowledge reference source generated by the model to keep consistent with the knowledge reference source in the truth value label, and the fact consistency is used for encouraging the answer content generated by the model to keep consistent with the semantics of the knowledge fragment corresponding to the generated knowledge reference source. By performing reinforcement learning training on the large model according to the reward function, the accuracy and reliability of the generated content of the large model can be improved, and the illusion problem of the large model can be effectively relieved.

The invention relates to an input adaptive precision Cordic algorithm implementation method and device. The method comprises the following steps: obtaining truth values of input x and y and comparing the truth values; performing most significant bit estimation on the truth value with the larger truth value, and obtaining a corresponding shifting mode; shifting the truth values of the input x and y based on the queried shifting mode; and inputting the complemented input x and y into the Cordic operation unit. According to the invention, the spare bit width in the input data is fully utilized to reduce or avoid the occurrence of error iteration, so that the occurrence of error iteration can be reduced to the greatest extent and the accuracy of the output result can be improved.

The present invention discloses a crowdsourcing labeling truth value reasoning method, device and storage device. The method is specifically as follows: construct a data set corresponding to each crowdsourcing worker; for a worker, if the worker has annotated a target instance, calculate the distance between each instance in the data set corresponding to the worker and the target instance, otherwise directly set the worker's labeling quality on the target instance to 0; sort the distance between each instance and the target instance, and find the K neighbor instances with the closest distance to the instance in the data set; use the K neighbor instances to measure the deviation and variance of the worker on the target instance; linearly combine the deviation and variance as the worker's labeling quality w on the target instance ir ; Repeat the above steps until the marking quality of all crowdsourcing workers on the target instance is estimated; select the marking of the worker with the highest marking quality from all workers as the inferred marking of the target instance. The beneficial effects of the present invention are: simple and easy to implement, high marking quality, and solve the problem of low performance of the truth value reasoning method.

The embodiment of the present application discloses a logic circuit truth table determination method, device and electronic equipment, relates to the technical field of electronic circuits, and can quickly extract the truth table of a logic circuit. The method comprises the following steps: based on a first group of input truth values in an input truth value list, input ports of a logic circuit described by a netlist are valued; the input truth value list is generated in advance according to the input ports of the logic circuit described by the netlist; based on the valuation of the input ports, the level value of the output ports of the logic circuit described by the netlist is calculated; it is judged whether the output ports have level values calculated; if the output ports have level values calculated, the calculated level value of the output ports is taken as the output truth value corresponding to the first group of input truth values. The present application is suitable for extracting the truth table of a logic circuit.

This application discloses a method and related apparatus for grading math problems, relating to the field of artificial intelligence technology. This application converts user responses in natural language into proof skeleton data in a pre-defined formal language, and uses a compiler of the pre-defined formal language as a deterministic oracle. By capturing compilation feedback signals, it determines the final evaluation result of the reasoning validity of the problem-solving process. By using the compiler's strict execution of substitution probability prediction, it provides deterministic logical truth values ​​for fuzzy natural language reasoning. Therefore, it can improve the reliability of the evaluation of the reasoning validity of students' problem-solving process.

The invention belongs to the technical field of knowledge questions and answers, and discloses a large model question and answer method and device based on a logic expression and a knowledge graph. The method comprises the steps of extracting a to-be-queried question from a text input by a user by using a large language model; inputting a to-be-queried question into the trained logic extraction and conversion module to generate a true value expression; inputting the truth value expression into the knowledge graph for query to obtain an effective truth value expression and a corresponding query result; and inputting the effective true value expression, the query result and the to-be-queried question into the large language model to generate a final answer. Through a small-scale large language model, questions and answers with high credibility and good interpretability can be provided for users.

The invention relates to a MC / DC coverage rate determination method and device based on a virtual machine and terminal equipment. The method comprises the following steps: analyzing an application source code file to obtain all conditional statements and conditional information, and sorting and simplifying the conditional statements and the conditional information to obtain an independent conditional list and corresponding conditional statement information; analyzing the executable target code file to obtain the address of the conditional statement and the symbol table information of the application program; adding an MC / DC coverage rate recording task in a virtual machine running environment, reading the address of each conditional statement by the task, calculating the truth value of each independent conditional simple logic expression and the truth value of the whole conditional judgment expression when the virtual machine runs to the address, and recording the coverage rate of each conditional statement; recording a file name, a function name, a line number, a judgment expression and a current execution truth value list result of the current statement into a specific file; and analyzing the record file after the operation of the virtual machine is finished, and calculating to obtain the MC / DC coverage rate of the conditional statements in the execution process and the MC / DC coverage rate of all conditional statements in the source code file of the application program.

A cause-effect graph generation device according to one aspect of the present disclosure comprises: a requirement analysis unit that uses a specification in which a requirement relating to a development target is described in natural language as input and uses a large language model to generate a condition relating to business logic represented by the specification and a truth table representing a relationship between the authenticity of the condition and the authenticity of results related to the business logic; a logical expression generation unit that uses the condition and the truth table as input and uses the large language model to generate, for each of the results, a logical expression which derives the results from the condition; and a cause-effect graph generation unit that uses the logical expression as input to generate a cause-effect graph which visualizes a relationship between the condition and the results.

The disclosed embodiments relate to a method and apparatus for constructing a distance truth library, a method and apparatus for detecting high-precision maps, an electronic device, and a storage medium, wherein the method for constructing the distance truth library comprises: obtaining map data, the map data being collected along the entire road by a collection device that meets set accuracy conditions; performing semantic segmentation based on the map data to obtain at least two first map elements; obtaining the distance truth value between two first map elements that are within an accuracy benchmark range among the at least two first map elements; and constructing a distance truth library based on the distance truth value between the two first map elements that are within the accuracy benchmark range. The disclosed embodiments can cause the distance truth library to include the distance truth value between any two first map elements that are within the accuracy benchmark range on the entire road section, thereby improving the accuracy of the map accuracy assessment when the map accuracy of the high-precision map is subsequently detected based on the distance truth library.

The invention discloses a formal verification method and system in an industrial control system. The method comprises the following steps: determining a target business entity corresponding to an operation object in a current message in the industrial control system; obtaining at least one target rule corresponding to the target business entity from a verification rule base; executing a logic verification operation on the operand value in the current message based on each target rule; if the judgment result of the logic verification operation shows that the target rule is violated, a safety response action is triggered; wherein the execution process of the logic verification operation comprises the following steps of: generating an instantaneous atomic proposition truth value corresponding to an operation value in a current message; obtaining truth values of other atomic propositions in the target rule from the system state cache to obtain a historical atomic proposition truth value; combining the instantaneous atomic proposition truth value and the historical atomic proposition truth value to obtain an atomic proposition truth value vector of the current message; and obtaining a judgment result corresponding to the true value vector of the atomic proposition, and defending logic attacks which utilize legal instructions but violate the technological process.

The present disclosure relates to a task planning method and device based on a large language model, an electronic device, a storage medium and a computer program product. The method comprises: for each task in a plurality of robot tasks, randomly sampling target parameters from a parameter set corresponding to the task; generating a target task instruction in natural language form based on the target parameters; constructing a structured step true value sequence corresponding to the target task instruction; inputting the target task instruction into a pre-trained large language model; comparing a structured step prediction sequence with the structured step true value sequence; and based on the comparison result, evaluating the inference performance of the large language model. In this way, since the present disclosure can automatically and quantitatively evaluate the structured inference ability of the large language model under fixed task constraints, the adverse effects of human factors on the inference performance evaluation of the large language model caused by relying on artificial labeling or subjective judgment can be effectively avoided.

Provided in the embodiments of the present application are a large-model performance assessment method and a related device. The method comprises: acquiring structured text description of a test material, wherein the structured text description comprises information of the test material in a plurality of dimensions; inputting the structured text description and a test question into a large language model, so as to acquire an evaluation true value; inputting the test material and the test question into a large model under test, so as to acquire an answer under test of the test question; and inputting the evaluation true value, the test question and the answer under test into the large language model, so as to acquire an assessment result for the answer under test.

The invention discloses a class-level object grabbing method based on an equivariant grabbing pose field. The method solves the problems that an existing method is low in reasoning speed, poor in pose generalization and the like. The invention provides an equivariant grabbing pose field which comprises four parts. Firstly, a grabbing pose field is constructed, object shape point cloud serves as a condition, class-level shape features are extracted through a neural network, and mapping from any input pose to an object feasible grabbing pose is learned. Secondly, training a grabbing pose field, quantifying a grabbing pose data set of an object by using L1 norm loss, and supervising the difference between a predicted pose and a grabbing pose true value; then, on the basis of reasoning of a grabbing pose field, grabbing pose prediction is used as guidance for updating a grabbing pose estimated value, and interpolation is carried out in the Lie algebra space; and finally, constructing a grabbing pose field to be equal, so that the network can be generalized to any object pose through the property of the construction instead of data enhancement.

Provided is a secret logical calculation device in a secret logical calculation system comprising three or more secret logical calculation devices for calculating ternary logic in a secret manner, wherein: a truth value of the ternary logic is expressed using two bits; the first bit x0 from between the two bits (x0, x1 used to represent the truth value of the ternary logic is a flag indicating whether the truth value is Undefined, and the second bit x1 represents a true / false value which is the truth value when the first bit x0 indicates that the truth value is not Undefined (however, when the first bit x0 indicates that the value is Undefined, the second bit (x0, x1) represents Undefined).

The invention discloses a method and a system for converting an RSML-e model into a Luter synchronous data stream language. The method is used for converting the RSML-e model into a Luter related type, a Boolean expression, an and-or table, a variable, a macro and a function. Type conversion processes undefined values by creating enumeration and structures. The Boolean expression conversion comprises truth value judgment, logical negation, pre-state reference and equivalence judgment. And constructing sub-expressions according to columns in the and-or table conversion, and using logic or connection. The variables are converted into a Luster node, and the ordinary variables and the input variables are used for processing initial values and conditions respectively. The macro and function are converted into nodes with input and return values, and the contained logic is converted into a Boolean expression. In the whole process, the mapping from the RSML-e logic to the Luter data stream is realized, so that the security attribute of the system can be verified by using a model detection tool, and the security of the system is improved.

Disclosed are a method and system for redundancy switch between two controllers for supplementary damping of a governor, relating to the technical field of control methods. The method includes acquiring a participation condition according to an output result, listing the participation condition in a truth value table and supplementing correlation logic; outputting the correlation logic as an image language, and simplifying the image language; and acquiring switch logic by a simplified image language, and transforming the switch logic into a programming language of the controller. A complex control relation can be decomposed in the disclosure, so that the difficulty can be reduced, and a correct and comprehensive control purpose can be achieved. Moreover, the method has universality.

The application discloses an automatic driving data set construction system and method, and belongs to the technical field of automatic driving; the system comprises a hardware and a data layer for collecting relevant data of a target scene and clock aligning the data; a perception and planning layer connected with the output end of the hardware and the data layer, for processing the aligned data, and outputting physically credible three-dimensional geometric true values, kinematic true values and expert planning trajectories; and a semantic cognition layer connected with the output end of the perception and planning layer, for key frame screening, fact dictionary construction and counterfactual reasoning, for converting numerical true values into cognitive annotation data, and obtaining an automatic driving data set, wherein the system is combined with the corresponding method to establish a three-in-one multi-paradigm unified architecture, which not only provides a high-precision three-dimensional geometric annotation for a modular model, and a closed-loop planning trajectory for an end-to-end model, but also innovatively introduces a counterfactual reasoning and hallucination suppression mechanism.

The invention relates to a method, particularly a computer-implemented method for determining a fatigue value of a person, the method comprising the steps of:i) displaying of:a) a first group (1) of symbols and a second group (2) of symbols,b) a selection rule (3) associating each of the symbols of the first group (1) of symbols to one symbol of the second group (2) of symbols,ii) highlighting one symbol (11) of the first group (1) of symbols,iii) receiving a user input from the person selecting one symbol (12) of the second group (2) of symbols,iv) determining a truth value based on the selected symbol (12) of the second group (2) of symbols, wherein the truth value assumes a true value if the selected symbol (12) is selected in accordance with the selection rule (3) and wherein the truth value takes on a false value if not, wherein the highlighted symbol (11) of the first group (1) of symbols is highlighted for a predetermined time based on a statistical response time of the person, and wherein if the user input is not received within the predetermined time in step iv), the method is set forth with step v) having the truth value set to the false value, or the method is set forth with step vi),v) storing the truth value on a non-transitory memory device,vi) changing the selection rule (3), such that each of the symbols of the first group (1) of symbols is associated anew to one symbol of the second group (2) of symbols,vii) changing the highlighted symbol (11) of the first group (1) of symbols,viii) repeating steps iii) to vii) at least once andix) based on at least some of the stored truth values, determining a fatigue value of the person.

The present invention relates to the technical field of crowdsourcing systems, and specifically to an anti-plagiarism iterative truth value discovery method based on lightweight homomorphic encryption. By adding the attribute of the working end weight in the truth value calculation process of the crowdsourcing task, the truth value is inferred according to the attribute of the working end weight to obtain a preliminary set of truth value answers, and whether the working end is the original author is determined based on the preliminary set of truth value answers. Based on the answer set of the original author, the plagiarist is determined by the similarity threshold, and the weight of the plagiarist is reduced. The next round of truth value calculation is performed according to the changed weight of the working end until the convergence criterion is met to output the final set of truth value answers. The crowdsourcing system can automatically monitor and automatically exclude plagiarism in the process of truth value calculation, thereby improving the accuracy of the final truth value result.

The invention relates to the technical field of water conservancy and hydropower engineering safety monitoring, and discloses a safety monitoring data abnormal value identification method based on a sliding window and morphological logic, and the method comprises the steps: processing an original data sequence through a window sliding at a fixed odd length, calculating a median and a median absolute deviation, and estimating a standard deviation; generating a dynamic upper and lower threshold sequence; comparing with a dead zone tolerance threshold value, converting into a logic vector, identifying a connected region with continuous truth values, and secondarily marking all data points in the region with the length exceeding the maximum allowable value as abnormal to obtain an abnormal value logic sequence; and performing linear interpolation processing on the data points marked as abnormal values in the abnormal value logic sequence to obtain a cleaned data sequence without abnormal values. According to the method, more than 95% of invalid data can be efficiently eliminated while the integrity of real signals is reserved, a high-reliability data basis is provided for data analysis and structural safety evaluation, and the method has important engineering application value.

System for causal-logical validation of AI output values, comprising an AI module and a logic validator, wherein the logic validator has access to an axiomatic reference database (Commons architecture).

The invention discloses a multi-modal data generation method and system for failure event injection in a complex scene, and belongs to the technical field of computer data generation. The method comprises the following steps of: 1, constructing a failure action and a recovery action, including obtaining and expanding an original script and defining and realizing a regularized failure type; 2, generating multi-modal data based on the failure action and the recovery action constructed in the step 1; and step 3, based on the multi-modal data generated in the step 2, forming a data set which can be directly used for VLMs failure perception evaluation through quality control, template design, truth value labeling and data division. The method is used for constructing a multi-modal reasoning data set, the data set comprises various failure events and corresponding recovery actions, and evaluation data in a video question-answer pair form is constructed based on multi-modal data and annotation files.

The present invention discloses a multi-truth value discovery method based on attribute fusion, which is an unsupervised multi-truth value discovery method. Considering the influence of auxiliary attributes on multi-truth value discovery, the data source expertise and consensus obtained by the auxiliary attributes and the activity of multi-truth attribute values are fused to obtain the support degree of the data source for the attribute value. The multi-truth value discovery problem is converted into a classification problem, and a neural network is used to automatically learn the underlying data features between the data source and the attribute value, so as to finally predict multiple truth values. The accuracy of the experimental results of the multi-truth value discovery method in the field of data integration can be improved. In the face of a multi-source noisy and conflicting data environment, the method can automatically identify, judge and infer complete truth values or data close to the truth values, thereby assisting users in making decisions and improving decision-making efficiency and quality.

The invention relates to a resource-optimized dynamic crowd sensing truth value discovery method, and belongs to the field of crowd sensing and crowdsourcing data management. The method comprises the following steps: obtaining initial capability estimation values of users through initial golden standard problem tasks, sorting the initial capability estimation values, and distributing common perception tasks for the users; and performing iterative calculation on perception data of all users participating in the same common perception task to obtain final truth value estimation of the current common perception task, thereby calculating a temporary capability index of each user, constructing a state vector and inputting the state vector into a reinforcement learning model to obtain a decision of a golden standard problem calibration task, and finally obtaining a standard problem calibration task. A gold standard problem calibration task is allocated to the user; and generating a reward signal based on the task release cost to update parameters of the reinforcement learning model, and optimizing generation of a decision. The objective of the invention is to solve the technical problems that in the prior art, a task allocation strategy cannot be dynamically adjusted, and resource waste or insufficient calibration precision is caused by a fixed-frequency GSQ publishing mode.

The embodiment of the invention provides a large model performance evaluation method and related equipment, and the method comprises the steps: obtaining a structured text description of a test material, and the structured text description comprises information of multiple dimensions of the test material; and inputting the structured text description and the test problem into the large language model to obtain an evaluation truth value. Inputting the test materials and the test questions into the to-be-tested large model to obtain to-be-tested answers to the test questions; and inputting the evaluation truth value, the test question and the to-be-tested answer into the large language model to obtain an evaluation result of the to-be-tested answer.

The invention discloses a basic large model truth value generation system, and the system comprises a system construction and initialization module which is used for carrying out system architecture design, model selection and training, and data preprocessing; the open vocabulary and cross-modal annotation module is used for performing user input analysis and dynamic annotation generation; the multi-modal feature fusion and semantic segmentation module is used for carrying out feature extraction and feature fusion; and the dynamic scene analysis and context understanding module is used for carrying out time sequence modeling and context analysis. According to the method, an innovative, flexible and efficient solution is provided for multi-modal data labeling through the advantages of efficient multi-modal fusion, open vocabulary support, automatic labeling process and quality control. Through wide application scenes and high-quality output, the method becomes an ideal technical choice in high-demand fields such as automatic driving and robot perception.