18 results about "Negative response" patented technology

A system may receive first reference information. The system may generate synthetic conversation records using a first generative artificial intelligence (AI) model. The system may generate response pairs based on the synthetic conversation records, where each response pair includes a respective positive response in accordance with the first reference information and a respective negative response in disaccord with the first reference information. The system may train a first set of generative AI parameters based on the plurality of response pairs and may merge the first set of generative AI parameters with a second set of generative AI parameters to generate a merged set of parameters. The system may receive a query and may provide a response generated by generative AI based on the merged set of parameters.

System and methods for generating training data for fine tuning a model in a data scarce setting are disclosed. In some embodiments, a disclosed method includes: receiving, from a user interface, an indication of a customer's interaction with a website, the indication including a customer query, generating, using a first model, a plurality of synthetic queries, generating, using the first model, a plurality of synthetic responses to each of the plurality of synthetic queries, the plurality of synthetic responses being generated based on one or more rules and each synthetic response of the plurality of responses having a positive response and a negative response, aggregating, using a second model, the plurality of responses to generate training data, and training a third model using the training data, the third model configured to generate a response to the customer query in real-time.

The invention provides a preference data synthesis method and device, and relates to the technical field of preference data synthesis, and the method comprises the steps: carrying out the processing of a positive response and a negative response of a preset instruction set under the action of each basic criterion based on a large language model, and determining the direction information under the action of each basic criterion; obtaining a vector representation of each diverse instruction in the diverse question instruction set under the action of each basic criterion based on a large language model; according to the vector representation and direction information of each diversified instruction under the action of each basic criterion, determining a target instruction set meeting the consistency requirement of the to-be-trained target model; on the basis of direction information corresponding to the target basic criterion, vector representation of each target instruction in the target instruction set under the target basic criterion is regulated and controlled, and preference data of the to-be-trained target model is synthesized; the technical problems that response needs to be output for multiple times and target model characteristics cannot be fully matched due to the fact that large language model output has randomness are solved.

The invention relates to a big language model training method and computer equipment. The large language model training method comprises the steps of obtaining a seed data set and a security rule set; the seed data set comprises abnormal user questions, positive response samples corresponding to the abnormal user questions and negative response samples corresponding to the abnormal user questions; obtaining a first positive reasoning sample and a negative reasoning sample based on the initial large language model according to the seed data set and the security rule set; and based on the first positive reasoning sample, the negative reasoning sample and the seed data set, constructing a first preference data pair, and training the initial large language model to obtain a trained large language model. By the adoption of the method, the problem that in the actual use process, when a user requires a large language model to output harmful content through prompt word guidance or external knowledge injection, the large language model cannot discriminate the intention of the user and cannot output the harmful content can be solved.

A method for authentication of non-fungible tokens (NFTs) includes receiving a validation request for an NFT from an external platform including a smart contract identifier; calculating an authentication score for the NFT based on at least (i) wallet activity associated with a creator blockchain wallet identified based on the smart contract identifier, and (ii) a fraud score of the external platform; and if the calculated authentication score for the NFT is below a predetermined threshold value, transmitting a negative response message to the validation request indicating unsuccessful validation of the NFT, or if the calculated authentication score for the NFT is above the predetermined threshold value, transmitting a positive response message to the validation request indicating successful validation of the NFT and instructing the external platform to execute a smart contract associated with the smart contract identifier.

A method for displaying digital media objects and dynamically calculating their rating values. A first digital media object is displayed on a first computing device. A first user provides one of predefined inputs corresponding either to a positive or a negative response to the first digital media object. The rating value of the first digital media object is increased if the response is positive and decreased if the response is negative. The numerical change of the rating value is calculated based on a coefficient value associated with the first user. The rating value of the first digital media object and the coefficient value of the first user are dynamically recalculated based on subsequent responses of the first digital media object from other users. The updated coefficient value is used to calculate subsequent numerical changes in the ratings of other digital media objects based on the first user's subsequent responses.

A method for verifying, by a receiver during a call, an alleged affiliation of an initiator with an organization comprises: receiving an initiator request comprising one or more first identifiers; generating a verification secret in response to the initiator request; sending the verification secret to the initiator; storing the verification secret in a database; receiving a receiver request to verify an alleged verification secret, the receiver request comprising one or more second identifiers; if there is a match between the second identifier(s) and the first identifier(s), comparing the alleged verification secret to the verification secret; and sending to the receiver: a positive response if there is a match between the alleged verification secret and the verification secret; or a negative response if there is no match between the second identifier(s) and the first identifier(s) or there is no match between the alleged verification secret and the verification secret.

Various methods and processes, apparatuses / systems, and media for questionnaire data digitization and reconciliation are disclosed. A processor generates an autonomous program for continuously monitoring shared mailbox for unread emails having questionnaire data containing a plurality of line items filled out by a client; converts, by utilizing an OCR tool, the questionnaire data containing the plurality of line items into a machine-readable format data; reads, by utilizing an automated reconciliation tool, the machine-readable format data for each line item; compares, by utilizing the automated reconciliation tool, data for each line item against a corresponding predefined guidance data; identifies, based on comparing, missing response data, negative response data, and insufficient response data corresponding to the questionnaire data filled out by the client by applying predefined rules; and automatically reconciles the missing response data, negative response data, and insufficient response data.

The invention discloses an information adaptive presentation method based on AR glasses and related equipment, and relates to the technical field of augmented reality. The technical means of evaluating the fatigue state of the user on the basis of cross-device use data, performing emotion recognition in combination with facial expression data, counting negative emotion feature events, weighting and calculating the content acceptance index on the basis of the negative degree and triggering information to simplify the display switching operation are adopted; therefore, when the screen use intensity of the user exceeds the personal historical average level, the device can actively monitor the emotional state change of the user, and the acceptance degree of the user to the current information content is evaluated by quantifying the occurrence frequency and duration of negative emotions. The complexity and density of information display are simplified when it is detected that the user generates a negative reaction, the problem of visual fatigue caused by the fact that AR glasses depend on a fixed watching duration triggering mechanism in the prior art is effectively solved, and then the technical effect of adjusting an information presentation strategy according to the real-time state of the user is achieved.

The invention provides an information transmission method and apparatus, and a sensing device. The method is applied to a first device, and comprises: sending first information for a sensing signal to a second device, the sensing signal being sent by the second device, the first information comprising at least one of the following items: a positive response; a negative response; sensing data, wherein the sensing data is obtained by analyzing the sensing signal; identification information of the first device; and identification information of the second device. According to the scheme of the invention, the second equipment can obtain the transmission condition of the sensing signal, so as to better control the transmission effectiveness of the sensing signal.

The invention discloses a cooking process-based cooking machine segmented feeding control method, and relates to the technical field of intelligent cooking control, and the method comprises the steps: building a feeding transaction, executing bidirectional micro-motion detection on a channel, evaluating the health degree of a weighing signal, and outputting a gating conclusion and a risk type; feeding according to a short coding beat pulse, collecting a single weighing signal and synchronizing related calculation to obtain a feeding amount estimated value and calculation confidence, stopping when the standard is reached, and supplementing feeding when the standard is insufficient; when the estimated value of the putting amount is stagnated or the confidence coefficient is reduced, fault typing is carried out based on an equivalent resistance index, a forward and reverse response asymmetry index and the consistency of the actuator, and the typing confidence coefficient is given; and implementing self-recovery or loss stopping according to the types, rechecking resistance falling, and submitting a transaction and recording a log after the delivery amount is increased and the confidence reaches the standard. Single-weighing verifiable metering is realized, misjudgment shutdown is reduced, and product consistency and continuous operation are improved.

This information processing method: collects information of at least one of verbal information or non-verbal information about one or more users; determines whether a negative reaction is detected from the user on the basis of at least one of the verbal information or the non-verbal information; and when the negative reaction is detected, reduces the negative reaction by controlling parameter changing devices that can change parameters of a space in which the user is present.

The application provides a 2E service and 31 service compatible adaptive processing method and system, and the method comprises the following steps: configuring at least one configuration entry, the configuration entry comprises an identifier, a service type, a session mode, a required security level and a service function pointer set; the service type comprises a 2E service or a 31 service; receiving and analyzing a UDS service request to obtain a to-be-processed identifier and a to-be-processed service type; judging whether there is a matched target configuration entry based on the to-be-processed identifier and the service type; if not, returning a negative response; if yes, performing session verification and security verification; if the session verification or the security verification fails, returning a negative response; if the session verification and the security verification both pass, calling a corresponding service function in the service function pointer set; and constructing a UDS response based on the return value type of the service function. The application realizes compatible adaptive processing of 2E and 31 services.

The invention provides a reward prediction method and device, electronic equipment, a computer readable storage medium and a computer program product. The method is applied to artificial intelligence. The method comprises the steps of obtaining a first cue word sample, a positive response text corresponding to the first cue word sample and a negative response text corresponding to the first cue word sample, and performing reward prediction on the first cue word sample and the positive response text as well as the first cue word sample and the negative response text through a teacher model, the method comprises the steps of obtaining first reward information and second reward information, training a to-be-trained student model based on a first difference between the first reward information and the second reward information to obtain a trained student model, and performing reward prediction on a target prompt word and a response text corresponding to the target prompt word through the trained student model. And obtaining the target reward information. According to the invention, the generalization ability of the student model can be improved.

The present application provides a multicast transmission feedback processing method, device, base station, and terminal. The method is applied to the base station and includes: the base station receives a first RLC status report sent by the terminal, wherein the first RLC status report is generated by the terminal based on data received via point-to-multipoint PTM transmission; if the status of the first RLC PDU in the first RLC status report is a negative response NACK, and it is determined to retransmit the first RLC PDU to the terminal via unicast, then based on the first status report, a second RLC status report is generated; wherein the status of the first RLC PDU in the second RLC status report is an acknowledgement response ACK; based on the second RLC status report, the RLC sending window of the PTM transmission is moved; and based on the first RLC status report, the first RLC PDU is retransmitted to the terminal via unicast. The embodiments of the present application can realize the reasonable movement of the RLC sending window under PTM, and ensure the smooth transmission of broadcast and multicast services to a group of terminals over the air interface.

The present disclosure relates to an electronic device and control method. The method includes obtaining situation information through a pre-defined situation recognition method; determining any one of a plurality of behavior patterns as a predicted behavior pattern on the basis of the situation information; generating a main query on the basis of at least one of the situation information and the predicted behavior pattern; outputting, on a display, a first virtual button corresponding to the situation information, a second virtual button corresponding to the predicted behavior pattern, and the main query; on the basis of receiving a negative response to the main query, activating the first virtual button and second virtual button; and on the basis of a user input on the first virtual button or second virtual button, updating any one of a mapping relationship between the situation information and the predicted behavior pattern, and the situation recognition method.

Embodiments of the present application disclose a message detection method and device, electronic equipment and medium. The method comprises: receiving a to-be-detected request message; performing full-cycle detection on the to-be-detected request message to obtain a message detection result; in a case where it is determined that the message detection result contains abnormal message data, determining priority configuration information of the abnormal message data; and generating target negative response information of the to-be-detected request message according to the priority configuration information of the abnormal message data. The technical solution of the embodiments of the present application can improve the accuracy of message detection, and further improve the portability and development efficiency of message detection.

The application discloses a false negative reaction determination method in automatic pure tone hearing threshold test, and the method comprises the following steps: obtaining initial hearing threshold test values of a target tester at multiple frequencies in a normal test process, wherein the initial hearing threshold test values comprise frequencies and sound intensities; selecting a set frequency and adjusting the sound intensity based on the initial hearing threshold test value at the set frequency; performing secondary test on the target tester according to the adjusted sound intensity to obtain second hearing threshold test values; comparing the initial hearing threshold test values with the second hearing threshold test values to obtain test difference values of the sound intensities between the initial hearing threshold test values and the second hearing threshold test values; and adjusting the test process according to the test difference values until the test result of the target tester is determined as a false negative reaction or the test is ended. The application can identify whether the target tester has a false negative reaction through the automatic test process, and the test process is adjusted according to the reaction of the target tester, so that the accuracy and reliability of the test are improved.