13 results about "Self correction" patented technology

Systems, methods, and apparatus to implement techniques to correct for jailbreak prompts input to generative language models are described. A prompt is generated that includes a jailbreak prompt, an original response to the jailbreak prompt, and a correction response. The generated response is then submitted to a generative language model and response returned.

The application discloses a hair detection method, device and equipment and a storage medium. The method adopts a YOLOV5 framework combined with KAPAO to perform target detection and key point detection on employees in a hair detection scene, obtains a pedestrian detection frame and corresponding pedestrian key point information, then uses a Self Correction for Human Parsing technology to analyze the human body of the pedestrians, obtains semantic segmentation information of each part of the human body, then uses a predefined rule to combine the key point information and the hair information to determine whether there is a hair behavior, and then uses an audible and visual alarm to perform real-time alarm. The method is real-time and effective, does not need to increase additional costs, has strong expansibility and generalization ability, and can be deployed in different hair detection scenes.

The invention provides a self-correction method and system for an AMR robot after position loss, and the method comprises the steps: judging whether a position loss event happens to the AMR robot, obtaining a first environment image in front of a current position of the AMR robot when the position loss event happens to the AMR robot, carrying out the object recognition of the first environment image to generate a first object list, and carrying out the self-correction of the AMR robot after the position loss event happens to the first object list; the first object list is a list composed of non-moving objects on the first environment image, obtaining the positioning credibility of each object in the first object list, and generating a second object list according to the positioning credibility, the second object list is composed of objects with the positioning credibility greater than a preset positioning credibility threshold in the first object list, determining two or more objects in the second object list as positioning objects, and positioning the AMR robot according to the coordinates of the positioning objects. And the AMR robot can be accurately and efficiently recovered and positioned after the position is lost.

The application discloses an automatic Meta analysis method and device based on large language model information extraction, relates to the field of Meta analysis, and comprises an analysis tool module, a memory module, a planning module and a Meta-Agent module; the Meta-Agent module is used for determining a PICO retrieval formula according to a target research theme, calling a thinking chain tool to perform analysis task decomposition; for any analysis task, corresponding tools in the analysis tool module are called to execute corresponding analysis tasks; in the process of executing the analysis tasks, long-term memory tools and short-term memory tools are called to store and call task analysis data; after the analysis tasks are executed, self-evaluation tools and self-correction tools are called to optimize decision-making; after the analysis task chain is executed, Meta analysis calculation is performed according to extracted control group data and intervention group data. The application can automatically and intelligently process complex Meta analysis tasks.

This invention discloses a clinical diagnostic method and apparatus for diseases, belonging to the field of disease diagnosis technology. The method can adaptively adjust the diagnostic path according to the specific circumstances of a case. It assesses the complexity of the condition through an intelligent agent and selects the most appropriate diagnostic procedure accordingly, thereby improving the accuracy and efficiency of diagnosis. By constructing a multi-agent system to simulate expert collaboration across different fields, it can integrate the opinions of multiple virtual experts, ensuring the comprehensiveness and reliability of diagnostic results and reducing the risk of misdiagnosis that may arise from a single model. The diagnostic process strictly follows clinical guidelines, and through multi-agent debate and self-correction procedures, it ensures that each diagnostic step is based on evidence, enhancing the interpretability of diagnostic results. This allows medical professionals to understand and trust the diagnostic output, facilitating its application in real-world clinical settings.

The application discloses a language model evaluation and text processing method and system, and a computing device, and relates to the fields of artificial intelligence and model evaluation. The method comprises the following steps: obtaining a text data set of a language model; determining evaluation indexes of the language model based on the text data set, wherein the evaluation indexes comprise indexes corresponding to at least one sub-ability of the language model, and the at least one sub-ability is an ability obtained by decomposing a self-correction ability of the language model according to different dimensions; and determining an evaluation result of the self-correction ability based on the evaluation indexes, wherein the evaluation result is used to represent whether the self-correction ability of the language model needs to be improved. The application solves the technical problem of low accuracy and efficiency of self-correction of a language model in the related art.

The high-redundancy safety protection and automatic correction diversion sheave comprises a diversion sheave body, a shaft pin, a counterweight frame and an automatic correction mechanism, a bearing seat is installed in the diversion sheave body, two sets of first rolling bearings are symmetrically installed in the middle of the interior of the bearing seat, and the bearing seat is installed in the middle of the shaft pin through the first rolling bearings; angular contact knuckle bearings are symmetrically installed in the space between the two ends of the bearing seat and a shaft shoulder of the shaft pin, the counterweight frame is fixed to the end of the shaft pin through cooperation of a bolt and a gasket, the self-correction mechanism is arranged on the outer side of the counterweight frame and used for correcting the angular deviation of the diversion sheave, and a distance sensor is arranged on the outer side of the counterweight frame. The distance sensor is located below the self-correcting mechanism, the angular contact knuckle bearing comprises an outer ring and an inner ring, and a gap is formed between the outer ring and the inner ring. It can be guaranteed that the inclination degree of the diversion sheave is within the set range, and the probability of accidents caused by inclination of the diversion sheave is reduced.

A radiation hardened flip-flop includes a plurality of secondary flip-flops. Each secondary flip-flop includes both a data input terminal and an alternate data input terminal. Each secondary flip-flop also includes an enable terminal that selectively enables use of the alternate data input terminal. The radiation hardened flip-flop includes an error detection circuit that detects whether an error is present in one or more of the secondary flip-flops and provides an enable signal to the enable terminals indicating the presence or absence of an error in one or more of the secondary flip-flops.

The invention provides a signal detection method, apparatus and device, and a storage medium, which can effectively avoid the problem of self-information interference without introducing iterative feedback and feeding back the estimation result of the symbol for self-correction by constructing a state based on an original received symbol and a known channel structure, and improve the detection efficiency of the signal. An interference structure with any complexity can be accurately described, high-precision symbol detection is realized, and the method is suitable for any time delay-Doppler extension scene. In the process of advancing from the state of one received symbol to the state of another received symbol, the differentiated undetermined sending symbols playing a key information updating role between adjacent states are concerned, all sending symbols do not need to be reprocessed, joint search for full combinations is avoided, and the transmission efficiency is improved. The calculation complexity is reduced from the index level to the polynomial level related to the difference scale between the states, and the processing capacity is improved on the premise that the detection performance is not reduced.

According to the AIS patient self-correction system and method provided by the invention, complete-cycle precise management of scoliosis is realized; through a 64-channel electrode array and a phase interference tactile feedback architecture, paravertebral muscle electrical activity can be captured in real time, subcutaneous precise stimulation can be realized, and the muscle imbalance detection sensitivity is obviously improved; according to the muscle imbalance index, a direct mathematical correlation between a muscle force difference value and a Cobb angle change is established, and the problem that biomechanical indexes and imaging parameters are disjointed in traditional evaluation is solved; based on a dynamic correction strategy of a PPO-Clip reinforcement learning algorithm, the training intensity can be automatically adjusted according to the real-time physiological status of the patient, so that the training compliance is further improved; the end-cloud cooperative processing architecture compresses key response delay to be extremely low, supports daily continuous pathological monitoring, realizes integrated closed-loop intervention, and remarkably reduces the development risk of the Cobb angle.

The application discloses an automatic driving end-to-end model self-correction method, device and medium. The method first evaluates multi-view images through an end-to-end automatic driving model, detects failure cases and extracts intermediate output information. Then, the causes of the failure cases are analyzed as multi-modal prompts, combined with bird's eye view layout and scene description, and data requirements are generated through a visual language model. Next, improved generation model Drive-Sora is used to generate multi-view video data with spatio-temporal consistency and / or diversity. Finally, the generated data is used as new training data to iteratively update and optimize the end-to-end automatic driving model, and through performance evaluation, it enters the next round of data generation and optimization cycle. The application significantly improves the robustness and generalization ability of the automatic driving model, reduces the collision rate and planning error in the long-tail scene, and provides strong support for the development and application of automatic driving technology.

The invention discloses a real-time simulation model automatic modeling method based on generative artificial intelligence, and belongs to the technical field of generative artificial intelligence and real-time simulation. Comprising the following steps: constructing a double-layer agent structure based on artificial intelligence, and configuring professional knowledge and a program database; inputting an instruction by a user to automatically generate an executable program, and performing self-correction; and finally, generating an executable program and an AI reply, and mapping the executable program and the AI reply to simulink through a mapping mechanism to realize automatic modeling. According to the method, a complete model is directly generated from a natural language, a user can input demand description, AI automatically analyzes demands, deduces logic and generates the complete model, and manual intervention is reduced. The method not only complements all sub-codes, but can generate codes of complete structures including function definition, input and output, parameter processing, core logic and the like from demands.

The invention relates to the technical field of electric arc welding track control, in particular to an electric arc welding track control system and method with a self-correction function, and the system comprises a welding equipment installation control main body which is provided with an installation frame; a driving mechanism and a regulation and control mechanism with a monitoring function are mounted in the mounting frame; the control method comprises the steps of welding part splicing and fixing, welding equipment multidirectional adjustment, welding seam real-time monitoring and rapid adjustment and alignment. The regulation and control mechanism with the monitoring function can upload shot pictures in real time, autonomously analyze the change condition of the position to be welded, and transmit the change condition of a welding seam into the welding equipment mounting control main body, so that the welding equipment mounting control main body knows the trend of the welding seam according to shot video images, and the welding quality is improved. By means of front-back adjustment of the automatic conveying belt assembly and adjustment of the welding assembly with the monitoring function, it is ensured that a welding part on the welding assembly with the monitoring function can always correspond to the position of a welding seam, and the purpose of automatic correction is achieved.