427 results about "Positive feedback" patented technology

A behavioral feedback system is provided to facilitate shaping a subject's behavior during an activity by obtaining user input. The user input is obtained pursuant to user-selection of a behavior-related representation from a set of behavior-related representations, for an activity of a subject, displayed on an electronic device of the user. Each representation corresponds to a different potential behavior of the subject during the activity, and the selected behavior-related representation relates to a user-observed behavior of the subject during the activity. The method further includes determining, by one or more processors, a positive or negative feedback indication associated with the selected behavior-related representation, and generating a feedback signal to send to an electronic device of the subject to indicate to the subject the selected behavior-related representation and the associated positive or negative feedback indication.

The invention discloses an anti-interference microwave electrically tunable filter regulation and control system and method, and particularly relates to the technical field of communication anti-interference, the interference suppression of microwave signals is realized through the following steps: firstly, based on a preset frequency range, monitoring the frequency spectrum change of the microwave signals in real time, and identifying the type and source of interference signals; when the microwave electrically tunable filter is initially tuned, a single-channel strategy is adopted, a preset suppression measure is executed according to interference frequency characteristics, and interference signals are effectively filtered out; finally, a feedback control mechanism is used, the performance of the filter is monitored in the adjusting process, the strategy is dynamically corrected, and the stability and high efficiency of the anti-interference performance are ensured; according to the method, the performance of the microwave filter is evaluated in real time through the environment change characteristic index and the environment complex characteristic index, the positive feedback or negative feedback signal is generated through the fuzzy reasoning model to serve as the basis for adjusting the work of the filter, and the response capacity to the interference environment is improved by adopting a feedback control mechanism; and the high-efficiency anti-interference performance is ensured.

Disclosed is a binarized neural network circuitry using a quasi-nonvolatile memory device. More particularly, the binarized neural network circuitry according to an embodiment of the present disclosure is characterized in that a diode structure is positioned in a channel region between a drain terminal and a source terminal, a gate terminal is positioned on the diode structure, an operation state is determined by occurrence of a latch-up or latch-down phenomenon due to a positive feedback loop in the channel region based on different voltages applied to the drain terminal and the gate terminal, respectively, and a plurality of quasi-nonvolatile memory devices that implement memory characteristics of remembering a memory state are included as holes or electrons in a potential well in the channel region due to the positive feedback loop are accumulated, and the plural quasi-nonvolatile memory devices are a pair of two quasi-nonvolatile memory devices that operate as a single synaptic cell and are connected in parallel to form an array circuit, memory states of the two quasi-nonvolatile memory devices are determined and stored based on an input line applied from an input line processor connected to the array circuit and a weight update signal applied from a synaptic line processor connected to the array circuit, and a MAC (multiply-accumulate) operation result is output using a combination of the memory states.

The invention discloses a double-fed wind power plant series compensation sending-out system subsynchronous oscillation suppression method for achieving negative damping full compensation design, and belongs to the technical field of impedance characteristic optimization, and the method comprises the steps: simplifying a double-fed wind power plant series compensation sending-out system, and obtaining the equivalent output impedance and series compensation line impedance of a double-fed fan; in the simplified system, the equivalent output impedance of the doubly-fed fan is further simplified, a negative damping factor index is introduced according to the simplified equivalent output impedance of the doubly-fed fan and series compensation line impedance, and whether subsynchronous oscillation exists in the system or not is judged; a positive feedback control branch for realizing negative damping full compensation is added in a simplified system, and positive feedback gain and a negative damping item are designed to realize positive and negative cancellation; and carrying out adaptive adjustment on the positive feedback gain under different working conditions. According to the invention, oscillation risk suppression can be realized, the drift of the oscillation mode under different working conditions can be suppressed, the subsynchronous frequency band impedance characteristic of the doubly-fed fan is effectively optimized, and the operation stability of the system is improved.

The invention relates to the field of curriculum personalized adjustment, in particular to a curriculum learning personalized adjustment method and system of an enabling system, and the method comprises the steps: obtaining user information and historical learning data of a user when a learning instruction of the user is received, and recognizing target curriculum data of the user according to the user information and the historical learning data; acquiring historical behavior data of the user, identifying a concentration attenuation trend of the user according to the historical behavior data, and segmenting the target course data into a plurality of course segments according to the concentration attenuation trend; when the user learns the curriculum segment, behavior data of the user is acquired in real time; judging the concentration feedback state of the user according to the behavior data; when the concentration feedback state of the user is positive feedback, generating a learning efficiency enhancement strategy and a composite excitation strategy; and when the concentration feedback state of the user is negative feedback, generating a cognitive load regulation strategy and a pressure balance strategy. The learning state of the user is judged according to the behavior data, and the system can flexibly adjust the learning strategy.

A comparator operating by receiving a supply of a first voltage (VDD / VSS) and a second voltage (VSS / VDD), the comparator comprising: a differential amplifier circuit (110) connected to a first power supply node (Np / Ns) that supplies the first voltage via a first and a second node (N1, N2), and which generates a first current (I1) and a second current (I2) in the first node and in the second node, respectively, wherein the first current and the second current have a current difference obtained by amplifying a voltage difference between an input voltage (Vin) and a reference voltage (VREF) predetermined between the first and the second voltage; a current source circuit (130) connected to the differential amplifier circuit via a third node (N3) between the first power supply node and a second power supply node (Np / Ns) that supplies the second voltage.connected in series and generating a third current (I3) which is a sum of the first and second currents at the third node; an output stage (170) connected between an output node (N6) and the first power supply node and having a current driving force to generate a first output current (I6p) proportional to the first current between the first power supply node and the output node; a current conversion circuit (140) having a current driving force to generate a second output current (I4n) proportional to the second current between the second power supply node and the output node; a minimum selection circuit (150) connected between a complementary output node (N5) and the first power supply node and causing the complementary output node to generate a current (I5) proportional to a minimum current of the first and second currents (I6p,I4n); and a latching circuit (160) connected between the output node and the complementary output node and the second power supply node, wherein the latching circuit performs a reset operation in which both the output node and the complementary output node are electrically connected to the second power supply node in response to a reset signal (RST), and performs a latching operation with positive feedback in which a change generated in the output node is amplified from the second voltage to the first voltage when a high-low relationship between the input voltage and the reference voltage is reversed from a reset cancellation time, while the output node and the complementary output node are electrically disconnected from the second power supply node during circuit operation after the reset cancellation.and sets the voltage at the output node to the first voltage, while the voltage at the complementary output node is set to the second voltage until the reset signal is next input.

The invention belongs to the technical field of industrial control methods and systems, particularly discloses a cold working process parameter optimization and plate rolling control method based on artificial intelligence, and relates to the technical field of cold working forming and intelligent control. The method comprises the steps that historical process parameters and corresponding plate quality data are collected, an evaluation function is constructed, and a process weight value is output; optimal parameters are screened according to the process weight and imported into cold working equipment; the curvature distribution of the plate in the plate rolling process is monitored in real time, and the plate rolling state is recognized; and executing a maintenance strategy or a correction strategy according to the state, and fusing correction feedback to optimize subsequent parameters. According to the method, the plate rolling precision and the forming consistency can be remarkably improved, the abnormal state self-adaptive correction capacity is achieved, dependence on manual experience is reduced, and the method is suitable for various plate rolling devices and plate types and has good universality and popularization value.

The invention discloses a peak code modulation circuit based on a dynamic memristor, which comprises the dynamic memristor and a signal conversion module, and is characterized in that the signal conversion module comprises a signal isolation module, a feature extraction module and a pulse conversion module which are connected in sequence. The dynamic memristor generates a voltage signal with a slow-rising and steep-falling characteristic under constant current input. The signal isolation module comprises a high-impedance input buffer circuit; the feature extraction module comprises an asymmetric time constant signal processing circuit used for performing enhanced extraction on the steep drop feature of the memristor output signal; and the pulse conversion module comprises a dual-threshold positive feedback hysteresis comparison circuit and is used for converting the output signal of the feature extraction module into a rectangular digital pulse. The circuit is specially designed according to the characteristics of the memristor, is simple in structure, is easy to integrate, can achieve the efficient conversion from an analog signal to a standard digital pulse, and is suitable for the fields of neuromorphic calculation and low-power-consumption signal processing.

The invention discloses a positive feedback self-supervised learning method based on characterization of self-distribution and clustering attributes, and belongs to the field of computer vision, and the method specifically comprises the steps: firstly, building a joint embedded architecture JEA composed of a data enhancement module, an encoder, a projection head and an optional prediction head; then, an input set D of given images randomly samples a group of images and inputs the images into a data enhancement module to obtain enhancement results, then the enhancement results respectively pass through an encoder to obtain corresponding encoding results, and then the enhancement results respectively pass through a projection head to obtain corresponding projection results; for the online network, based on an output result of the projection head, calculating an output result after passing through the prediction head; calculating and optimizing a cross entropy loss function on the basis of an output result of each module of the joint embedded architecture JEA so as to update parameters of the JEA, and finally performing performance evaluation on the trained model on a computer vision task; according to the method, inherent clustering attributes of model coding are utilized, and the model performance is further improved through positive feedback learning.

The invention discloses a low dropout linear regulator with a wide dropout range, which is characterized in that an error amplifier is supplied with power by output voltage of the low dropout linear regulator, and a common-gate buffer is supplied with power by input voltage of the low dropout linear regulator, so that a direct interference path of the input voltage to the output only exists in a common-gate buffer stage and an output stage; and the power supply rejection ratio is improved. The common-gate buffer solves the problem of margin loss caused by a traditional source follower circuit, a positive feedback loop arranged in the common-gate buffer is used for improving the slew rate of a driving signal VPGATE and providing a certain gain, and therefore the problem of margin under low dropout is solved, and the reliability of the driving circuit is improved. And the compensation circuit can adjust the compensation amount of the low-dropout linear regulator when the input voltage is higher than a preset threshold value, so that the stability of the low-dropout linear regulator under different working conditions is improved.

The invention relates to the technical field of integrated circuits, in particular to a band-gap reference voltage source circuit with high-order compensation, which comprises a PTAT (proportional to absolute temperature) generation circuit, a high-order compensation circuit and a pre-voltage-stabilizing circuit, wherein the PTAT generation circuit generates PTAT current, and the high-order compensation circuit performs second-order compensation on the PTAT current; the pre-voltage-stabilizing circuit is used for driving the PTAT generation circuit and the high-order compensation circuit and comprises a starting circuit, a positive feedback loop and a negative feedback loop; the gain of the positive feedback loop is lower than that of the negative feedback loop; according to the band-gap reference voltage source circuit, the temperature coefficient and the power supply rejection ratio of the circuit are remarkably optimized by integrating high-order temperature compensation, a pre-adjusting power supply and a high-gain feedback loop, and the designed band-gap reference voltage source circuit has the advantages of being extremely low in temperature drift and high in precision.

The invention discloses a timing method and device, equipment and a medium, and relates to the technical field of communication. The method is applied to a first electronic device providing a reference clock, the first electronic device is in communication connection with at least two second electronic devices, and the method comprises the steps that for any second electronic device, a timing instruction is sent to the second electronic device based on the communication connection with the second electronic device, the timing instruction is used for indicating the second electronic equipment to start a broadcast scanning function; broadcasting operation is executed, and the broadcasting operation comprises the steps that according to the current clock of the first electronic equipment and the identity label of the first electronic equipment, a time correction expansion data packet is generated, and the time correction expansion data packet is broadcasted in an expansion mode once; and determining that the timing is completed under the condition that positive feedback which is sent by the corresponding second electronic equipment and represents that the timing expansion data packet is scanned is received based on each communication connection. Based on the method, the first electronic device can control clock synchronization of different second electronic devices.

The invention belongs to the technical field of network security, and particularly relates to a log analysis method using an AI large model, which is applied to a network security monitoring platform and comprises the following steps: acquiring source data of a log file; calling an AI large model in the platform to generate a corresponding regular expression based on the source data; sending the regular expression to the user for verification so as to obtain a final regular expression verified by the user; analyzing the source data according to the final regular expression to obtain an analysis result; and mapping the analysis results of the source data to the corresponding field templates one by one to generate standard log analysis results. According to the method, the regular expressions of the log files are intelligently recognized through the AI large model, efficient and accurate standardized analysis of the log files is achieved, meanwhile, the AI large model is further optimized through the manual feedback result, positive feedback circulation is achieved, and the performance of the AI large model is continuously optimized.

The invention provides an electrostatic protection structure, a preparation method and an electrostatic protection circuit. The electrostatic protection structure comprises a substrate, a well region which is formed on the substrate and has a second doping type, a gate structure which is formed on the surface of the well region, and a drain region and a source region which are formed in the well region at two sides of the gate structure, the drain region at least comprises a first drift region with a first doping type, a first heavily doped region with the first doping type and a second heavily doped region with a second doping type, wherein the first heavily doped region and the second heavily doped region are formed in the first drift region at intervals; and the source region at least comprises a second drift region with the first doping type and a third drift region with the second doping type which are arranged at intervals, a fourth heavily doped region with the first doping type formed in the second drift region, and a fourth heavily doped region with the second doping type formed in the third drift region. A positive feedback loop is formed in the electrostatic protection structure, so that the amplified ESD current is discharged.

A power amplifier with improved stability, gain, flatness and return loss and reduced the size, which comprises a two-stage power amplifier in balanced cascode topology that consists of a first common-source stage feeding into a first common-gate stage; a second common-source stage feeding into a second common-gate stage; an input matching network for matching the input impedance of the a two-stage power amplifier to a desired signal source impedance; an output matching network for matching the output impedance of the a two-stage power amplifier to a desired load; a first positive feedback path between the gate of the first common-source stage and the drain of the second common-gate (CG) stage and a second positive feedback path between the gate of the second common-source stage and the drain of the first common-gate (CG) stage; a first negative feedback path between the drain of the first common-gate (CG) stage and the gate of the first common-source stage and a second negative feedback path between the drain of the second common-gate (CG) stage and the gate of the second common-source stage; a first inter-stage matching network between the first common-source (CS) stage and the first common-gate (CG) stage; and a second inter-stage matching network between the second common-source (CS) stage and the second common-gate (CG) stage.

The invention relates to the technical field of oscillators, and discloses a low-noise broadband voltage-controlled oscillator based on a cross coupling differential structure, which comprises a resonant cavity, a differential negative resistance circuit and a current bias circuit, and is characterized in that the resonant cavity adopts an LC resonant structure and is used for determining frequency selection and phase characteristics of the low-noise broadband voltage-controlled oscillator; the differential negative resistance circuit adopts a cross coupling differential pair structure formed by two bipolar junction transistors and is used for providing enough gain and positive feedback so as to generate a negative resistance effect; the current bias circuit uses a bipolar junction transistor to construct a bias circuit structure and is used for providing stable bias current and ensuring that the bipolar junction transistor in the differential negative resistance circuit works in an amplification region.

The invention relates to the field of user interaction technology, specifically a method and system for real-time feedback of live broadcast barrage based on interactive semantic matching. The method receives and caches user barrage in real time, performs density analysis and priority evaluation on the barrage stream in combination with filtering and resource optimization rules, dynamically adjusts the response frequency and filters key barrages; by introducing the intention unit of the distillation processing mechanism, extracts user intention labels from key barrages and generates vectors, and simultaneously obtains the historical behavior data of the sending user, integrates the intention label and vector with the current real-time scene data of the live broadcast room, and constructs a barrage vector; performs similarity calculation based on the live broadcast content and the user's historical behavior to obtain the first feedback content; generates the second feedback content based on the reinforcement learning sentiment analysis strategy and the user's preference memory; and embeds the second feedback content into the current live broadcast interface through the client. The present invention realizes real-time positive feedback and intelligent guidance of live broadcast barrage through interactive semantic matching.

The invention discloses a land freeze-thaw wind erosion simulation measurement device, which relates to the technical field of soil testing and comprises a testing box, and temperature controllers are symmetrically mounted at the top end of the inner side of the testing box. The functions of slope adjustment, displacement control and dynamic driving are integrated through linkage design, the frozen soil wind erosion testing efficiency and precision are remarkably improved, a slope adjustment mechanism is driven by a bevel gear set, a supporting frame is driven to move reversely through a reverse spiral groove, automatic adjustment of the inclination angle and orientation of the frozen soil slope is synchronously achieved, and manual procedures are reduced; the displacement mechanism drives frozen soil to perform horizontal displacement and vertical reciprocating motion through gear-fluted disc transmission, simulates dynamic interaction and dispersion erosion effects of a natural wind field, observes and reveals a wind erosion positive feedback mechanism in combination with micro-topography evolution, and improves data universality; the driving mechanism utilizes an electric push rod and a double-shaft motor to cooperatively control meshing switching of a bevel gear set and a movable / fixed gear, precise switching of gradient adjustment and a dynamic displacement mode is achieved, and testing continuity is guaranteed.

The invention discloses a complex system fault root cause analysis method and system based on a knowledge graph. The root cause analysis method comprises the steps of collecting fault data; the method comprises the following steps: retrieving similar document vectors and knowledge graph data, and fusing heterogeneous knowledge; integrating the generated heterogeneous fusion knowledge and fault data provided by a user into a reasoning text required by the LLM, and generating a report by the LLM through the generated reasoning text; scoring the report, and if the score is lower than a threshold value, triggering to regenerate the report; if the related engineer solves the corresponding fault according to the analysis report, forward feedback is carried out, and the analysis report and fault data are updated to the knowledge system; and if the scheme cannot solve the fault and the fault is solved manually, related engineers input the manually solved scheme into the scheme and update the knowledge system. The efficiency of the whole process is improved.

The invention discloses a DRU-MMC direct-current parallel sending-out system for offshore wind power and a control method, and aims at achieving high-economical-efficiency and high-efficiency electric power sending-out of deep and far sea wind power clusters. The system comprises a first offshore wind power plant based on a network construction type fan, a second offshore wind power plant based on a network following type fan, an offshore MMC converter station and an offshore DRU converter station. The offshore MMC converter station adopts a voltage-frequency control strategy added with active power-voltage amplitude negative feedback control and reactive power-frequency positive feedback control, the land MMC converter station adopts constant direct current voltage control, the first offshore wind power plant fan adopts network construction type control, and the second offshore wind power plant fan adopts network following type control. According to the direct current parallel sending scheme of the DRU and the MMC, the technical advantages of the MMC and the economical efficiency of the DRU are considered, the stability and reliability of the system are improved by adopting the specific control strategy, and the economical efficiency of the offshore wind power sending system is improved.

The invention discloses a user intention dynamic correction feedback system based on large model personalized recommendation, which comprises a fusion module, and the fusion module is used for constructing initial probability distribution of user intentions based on multi-modal interaction data. Multi-modal fusion and Dirichlet modeling are adopted, user intentions are captured in an omnibearing mode, the intention adjusting module is dynamically optimized according to the environment, intention understanding is made to be matched with scenes in real time and accurately meet user requirements, the aggregation module uses an attention mechanism, the importance of explicit and implicit feedback is automatically distinguished, multi-source information is efficiently integrated, and the user experience is improved. A reliable basis is provided for follow-up recommendation, key signals are focused, user preferences are comprehensively and efficiently grasped, a recommendation module is based on an accurately-corrected intention vector, exploration and utilization are balanced through a dobby machine algorithm, known preferences are met, potential interests are mined, the recommendation satisfaction degree is improved, and the recommendation efficiency is improved. Therefore, the adaptability and expansibility of the system are enhanced while the recommendation effect is improved.

The invention belongs to the technical field of integrated circuits, and particularly relates to a dynamic comparator suitable for a low-voltage high-speed analog-to-digital converter, the dynamic comparator adopts a three-stage latch circuit structure to realize rapid comparison of weak input signals so as to generate corresponding digital signals, and the method specifically comprises the following steps of: in a pre-amplification stage, pre-amplification is carried out; the positive feedback effect of the cross-coupled transistor pair is utilized to efficiently convert a tiny input voltage difference (lt; 10mV) into a recognizable voltage difference, and then through two-stage regeneration of the main latch module and the auxiliary latch module, a logic level is quickly output to complete quantization judgment. The auxiliary latch stage is additionally arranged on the basis of a traditional cross coupling latch circuit, the working speed of the comparator is effectively increased under the condition that only small area overhead is introduced, and the probability that the dynamic comparator has a metastable state is greatly reduced.

The invention discloses a low-trigger high-maintaining-voltage silicon-controlled electrostatic protection device and a manufacturing method thereof, an N trap is added to the right side of a cathode of a traditional double-trap silicon-controlled device to form three traps, the three traps are connected to an anode through N + injection, and compared with a traditional one-way low-trigger silicon-controlled electrostatic protection device, the low-trigger high-maintaining-voltage silicon-controlled electrostatic protection device has a shorter distance between the cathode and the anode. Besides, an NMOS (N-channel Metal Oxide Semiconductor) and P + injection are embedded in a P trap of the device, a grid drain of the NMOS is connected with the P + injection, a source of the NMOS is connected with a cathode, a surface ESD (Electro-Static Discharge) current discharge path is added on the basis of a traditional unidirectional silicon controlled electrostatic protection device, and the positive feedback effect of the SCR discharge path can be effectively inhibited. Therefore, the device has the characteristics of low trigger voltage and high maintaining voltage, and a core circuit of a chip is effectively protected from the risk of latching. The device can be applied to ESD protection of the I / O port with the working voltage of 0-5.5 V.

The embodiment of the invention provides a gas safety knowledge base updating method, device and equipment and a storage medium, and the method aims at a target model, based on a gas safety knowledge base and a target knowledge result corresponding to any safety scene problem output by real-time safety data, and based on a model processing process corresponding to the target knowledge result. Obtaining processing process data, and obtaining result feedback information corresponding to the target knowledge result; determining a performance evaluation value and an effect evaluation value based on the processing process data and the result feedback information; the performance evaluation value is used for representing the processing performance of the knowledge processing process corresponding to the target model, and the effect evaluation value is used for representing the effect of the target knowledge result output by the target model; and updating the gas safety knowledge base based on the performance evaluation value and the effect evaluation value. Through continuous positive feedback optimization of the gas safety knowledge base, the accuracy and timeliness of contents in the gas knowledge base are ensured, and the updating efficiency and updating accuracy of the knowledge base are improved.

The invention discloses a signal acquisition circuit for a brain-computer interface, and relates to the technical field of integrated circuits. The acquisition circuit comprises a sensing amplifier, wherein the sensing amplifier comprises a chopping modulation capacitance coupling instrument amplifier and a band-pass filter which are connected in series; the chopping modulation capacitance coupling instrument amplifier comprises an input main chopping modulator, an input capacitor, a negative feedback circuit, a first amplification circuit and a second amplification circuit which are sequentially connected in series; a pre-charging circuit, an input capacitor and a first amplifying circuit are connected to the input main chopping modulator in parallel, a positive feedback enhancing loop is connected to a second amplifying circuit in parallel, the first amplifying circuit and the second amplifying circuit are connected with a first direct-current servo loop and a negative feedback circuit in parallel, and a ripple suppression loop is connected to the second amplifying circuit in parallel. And the second amplification circuit is connected with the signal output end through the band-pass filter. The sensing amplifier in the acquisition circuit can output the electroencephalogram signals with higher quality, so that the quality of the acquired brain-computer interface signals is higher.

The invention relates to a training method of a content recommendation model, a content recommendation method and device and a vehicle, and the method comprises the steps: recognizing an original positive feedback sequence and an original negative feedback sequence in interaction data of a user for an item; inputting the original positive feedback sequence and the original negative feedback sequence into a large language model, wherein the large language model outputs positive feedback semantic representation, negative feedback semantic representation, user features and item features; eliminating noise interference of the original positive feedback sequence and the original negative feedback sequence according to the positive feedback semantic representation and the negative feedback semantic representation, and generating a data set according to the positive feedback sequence and the negative feedback sequence after noise interference elimination, the user features and the item features; and training the content recommendation model by using the data set. Therefore, the problem that in the prior art, content recommendation is carried out on the user based on positive feedback, the negative feedback effect is ignored, and the diversity of the recommended content is low is solved.

The invention relates to the field of data security, in particular to a network security intelligent detection method based on big data, and the method comprises the steps: collecting network flow data, a terminal system call sequence and a user operation behavior log, and carrying out the data fusion processing to generate a unified behavior event flow; selecting a key behavior event based on an information entropy threshold value, performing time alignment through a dynamic time warping algorithm, and constructing a behavior gene map containing a communication association gene, an operation sequence gene and a behavior time sequence gene; a dynamic behavior baseline model is established by using unsupervised learning, and gene mutation detection and alarm are realized by calculating the deviation degree of each gene dimension; the detection performance is evaluated based on the false alarm rate, model parameters are optimized through a negative feedback mechanism, and acknowledged attack features are stored in a sharable threat gene feature library through a positive feedback mechanism. According to the method, the detection accuracy is continuously improved through a closed-loop learning mechanism, and a self-adaptive safety protection system with self-optimization capability is constructed.

The invention provides a hysteresis comparator, and belongs to the technical field of electronics. The hysteresis comparator comprises an open-loop operational amplifier circuit and a positive feedback circuit; the positive feedback circuit comprises a switching tube device and a current source, the switching tube device is electrically connected with the open-loop operational amplifier circuit and the current source, and the positive feedback circuit is used for forming an internal positive feedback path with the open-loop operational amplifier circuit when the switching tube device is switched on and adjusting hysteresis voltage through the current source. According to the invention, the hysteresis voltage can be regulated through the current source, so that the stability of the hysteresis voltage is ensured.

An unsupervised open relation extraction text processing method based on interaction of large and small models comprises the steps that firstly, a large language model generates a preliminary pseudo relation label from an input statement, and important clues are provided for subsequent feature relation clustering; secondly, the pseudo labels are used for supervising and fine tuning of a small language model, probability distribution of each category is generated through feature learning of relation instances and a relation guidance feedback algorithm, and the pseudo labels are optimized; and finally, selecting a sample with a high confidence probability as a demonstration sample, and providing positive feedback for the large language model so as to reprocess instances with uncertain relationships to improve relationship extraction. The performance of the InstructORE model is superior to that of an existing non-supervision baseline model, and the InstructORE model has relation extraction performance equivalent to that of a weak supervision method in most cases. The method provides powerful support for construction of knowledge maps, knowledge questions and answers and other applications, and has wide application prospects and popularization value.

The invention relates to a gradient perturbation optimization method for implicit matrix decomposition based on differential privacy, and belongs to the field of computer recommendation systems and the field of information security. An implicit feedback matrix composed of 0 and 1 is used for representing feedback data of a user to an item, 0 represents negative feedback, and 1 represents positive feedback. Randomly initializing a user implicit factor matrix W and a project implicit factor matrix H; a data set and a target function L (W, H) are constructed, W and H are optimized and solved by adopting a matrix decomposition method based on Bayesian personalized sorting, and a result W is stored. Randomly initializing a noise adding item implicit factor matrix # imgabs0 # and an initial record matrix # imgabs1 # by using the trained W, substituting the W into L (W, H), constructing a target function # imgabs3 # about # imgabs2 #, calculating a gradient # imgabs6 # of # imgabs4 # about # imgabs5 #, cutting the gradient by using a cutting threshold C, and obtaining an optimal privacy budget size lambda t in each round of training by using an overall mse calculation formula, then, the gradient is disturbed by using lambda t and the sensitivity of the gradient to obtain # imgabs7, the # imgabs8 is aggregated to obtain a cumulative gradient, and # imgabs9 is used for updating an item implicit factor matrix by using the cumulative gradient and # imgabs10; # imgabs11 is repeatedly updated to obtain a final item implicit factor matrix # imgabs12, and the final item implicit factor matrix # imgabs12 is distributed to each user; and the user u calculates a predicted value # imgabs13 on the equipment of the user u according to the predicted value, sorts unobserved interaction items in a descending order, and finally selects the first K items from the interaction items to recommend to the user u. The method has good privacy protection and model accuracy.