31results about How to "Reduce forgetting" patented technology

The invention provides an intelligent automobile-key system based on bluetooth communication, which comprises a PEPS (Passive Entry & Passive Start) controller, a vehicle-mounted bluetooth module, an electronic terminal and a door-handle switch, wherein the PEPS controller is connected with the vehicle-mounted bluetooth module and the door-handle switch, the PEPS controller is used for controlling a vehicle antitheft system and an engine system, the vehicle-mounted bluetooth module is used for realizing bluetooth communication with the electronic terminal, the electronic terminal has an intelligent automobile-key application program, and the door-handle switch is used for the input of user unlocking and locking requests. Because the vehicle-mounted bluetooth module exists, a vehicle is controlled to actively send out a bluetooth signal for actively identifying and authenticating the electronic terminal, such as a mobile phone, so that complicated operation is avoided.

The invention relates to a prompting system for articles in a trunk of a taxicab and a control method thereof; the promoting system is used for reminding passengers to prevent the articles from being left in the taxi and comprises an instrument control unit; the promoting system also comprises a weighting sensor, a luggage monitoring unit and a prompting device, wherein the weighting sensor is used for detecting the weight of the trunk, the luggage monitoring unit judges whether the luggage exists in the trunk according to the variation of weight detected by the weighting sensor; the prompting device is used for reminding drivers and passengers of the luggage in the trunk; the luggage monitoring unit is connected with the instrument control unit; the weighting sensor is connected with theluggage monitoring unit; and the prompting device is connected with a meter control unit. The invention provides the prompting system for articles in the trunk of the taxicab and the control method thereof, wherein the prompting system has a detecting function, can give out pertinence prompts according to detecting results and solves the problem that the traditional prompting system gives out prompts without pertinence.

The embodiment of the invention discloses a schedule reminding method and a terminal. The trouble that a user needs to manually input a schedule reminding event in the prior art is solved. The method disclosed by the embodiment of the invention comprises the following steps: receiving communication content; judging whether the communication content comprises target information; if the communication content comprises the target information, generating a target event according to the target information; displaying the target event on a user display interface so as to receive an instruction input by the user to the target event; and determining the target event as the schedule reminding event when the instruction is used for indicating that the target event is the schedule reminding event.

The embodiment of the invention provides a collecting method and device of pages and a mobile terminal. The method comprises the steps that collection operations aiming at loaded pages in first applications are detected; according to the collection operations, page information of the pages is obtained; according to the page information, favorite information is aggregated in second applications. On one hand, the pages of different first applications are collected in the second applications, it is avoided that users log in accounts and collect the pages in different first applications, the number of accounts is reduced, the convenience of management processing is improved, the number of confused collected pages is reduced, and the first applications are prevented from being started repeatedly for searching; on the other hand, additional collection functions can be provided for the first applications which do not support the collection functions, so that page collection is achieved, additional recording by using modes such as bookmarks is avoided, and the probability of the collected pages being forgotten by the users is reduced.

The invention relates to an incremental learning method based on generative adversarial network knowledge distillation. The method comprises the following steps: S1, enabling a generator to generate an old pseudo training sample with a sample label according to input condition information, and combining the old pseudo training sample with a conventional new training sample, thereby forming a new training data set; S2, training a new generator and a discriminator on the original generative adversarial network according to the new training data set; S3, training a classifier based on the new training data set, training the classification capability of the classifier by utilizing the original label information of the sample in the training process, and guiding the learning of the classifier by taking a new discriminator as a teacher model to obtain a new classifier for generating a high-quality image. According to the method, the total category related information of the new category andthe old category is effectively provided for the classification model, so that forgetting of the classification model in incremental learning can be reduced, and classification precision is improved.

The invention discloses a cardiopulmonary resuscitation guider and guiding method based on an AR technology. The cardiopulmonary resuscitation guider comprises a camera shooting module used for obtaining first-aid site environment information and / or patient information, a virtual image processing module used for calling and / or constructing a corresponding virtual first-aid guidance video / image, adata analyzing and processing module further used for determining the display position of the virtual first-aid guidance video / image in the first-aid field environment according to the first-aid fieldenvironment information and / or the patient information acquired by the camera module, and an AR display module used for displaying the virtual first-aid guidance video / image in the first-aid field environment and / or displaying the virtual first-aid guidance video / image in an overlapping manner at a corresponding position. According to the invention, the situation of forgetting or negligence of first-aid personnel due to tension can be relieved, and first-aid guidance can be provided according to duty; for ordinary people who are not trained, the cardiopulmonary resuscitation first-aid operation can be carried out instantly.

The invention discloses an affair reminding method and device and belongs to the field of computers. The method comprises the steps of receiving affair reminding information sent by a first client side and second client side information, and sending the affair reminding information to a second client side according to the second client side information so that affair reminding can be carried out by the second client side according to the affair reminding information. The device comprises an information receiving module and an affair reminding module. According to the affair reminding method and device, the affair reminding information sent by the first client side is sent to the second client side according to the second client side information so that the affair reminding can be carried out by the second client side according to the affair reminding information, the automatic reminding function is achieved, and affair forgetting or losing probability of an affair executor is reduced.

The invention provides a deep reinforcement learning training method and a computer readable storage medium, and the method comprises the steps: assigning a situation number, and initializing a weight parameter of a deep reinforcement learning multi-head neural network model; carrying out stochastic decision by an intelligent agent and collecting samples and storing the samples in an experience playback buffer area; according to the situation number, adopting an online clustering algorithm to achieve self-adaptive situation division to obtain situation division and situation centers at the end of the current moment; randomly sampling samples from the experience playback buffer area, and distributing each sample to the closest situation; training a shared feature extractor and a weight parameter of a corresponding output head according to a situation corresponding to the sample, synchronously updating the weight parameters of other output heads in combination with knowledge distillation loss, and estimating a value function; and in the next time step, through the intelligent agent, continuing to make a decision according to the value function, making samples collected and stored in the experience playback buffer area, and repeating the steps until the pre-specified number of training times is completed or convergence is achieved. The stability and plasticity of model training are improved.

The invention discloses an operation authentication method and device, a storage medium and an electronic device. The method comprises the steps of acquiring a resource operation request sent by a target device, wherein the resource operation request is used for indicating an application client running by using a target account in the target device to request to execute resource operation; obtaining a first target password input by the target device under the condition that the target device is determined to be a binding device of the target account according to the resource operation request;under the condition that the first target password is matched with a verification password of a target account stored in a server, obtaining a first dynamic password dynamically generated for the resource processing operation by a token development kit in the target equipment; and under the condition that the first dynamic password is matched with a second dynamic password generated by the server, setting the permission of executing the resource operation by the target device to be allowed to be executed, the second dynamic password being used for verifying the resource operation. According to the invention, the reliability and portability of operation authentication can be improved.

The invention discloses a multifunctional portable earphone box. The multifunctional portable earphone box comprises a box body, wherein a storage groove is formed in the bottom of the box body; retractable rods are hinged to the left-side upper end of an inner cavity of the storage groove and the right-side lower end of the inner cavity of the storage groove; suction discs sleeve the bottoms of the retractable rods; the right side of the top of the box body is hinged with a box cover; a plane mirror is arranged at the top of the inner cavity of the box cover; a locking block is arranged at the left side of the bottom of the box cover; a semispherical groove is formed in the right side of the locking block; a button seat is arranged at the left side of the box body; a switching button is inserted into the left side of the button seat; an extruding block is fixedly connected with the switching button; locking springs are fixedly connected with the upper and lower ends of the right sideof the extruding block; a limiting stopping piece is fixedly connected with the right side of the switching button; a locking bulge is arranged at the left side of the limiting stopping piece. The multifunctional portable earphone box is reasonable in structure design and an earphone is convenient to store; the possibility of forgetting to carry an earphone box and a mirror is reduced; meanwhile,the earphone box is convenient to fix in a utilization process, so that the female can use the mirror conveniently and can make up conveniently, and the multifunctional portable earphone box is convenient to carry and use.

A locking and unlocking system includes a controller. The controller is configured to acquire a state of locking and unlocking, of a vehicle that is locked and unlocked based on locking and unlocking instruction information acquired from a portable terminal, and send a notification to a first device associated with the vehicle, when the vehicle is not locked during a period of time from when the vehicle is unlocked based on the locking and unlocking instruction information acquired from the portable terminal until a first timing.

The invention discloses an electronic work list system and a using method thereof. The electronic work list system consists of a data storage, a data processing server, at least one handheld terminal and a gateway, wherein an electronic work list, a data base query system and a wireless communication unit which is used for being contacted with the handheld terminal are preset in the data storage; the data processing server is electrically connected with the data storage, and a software program system for performing read, edition, modification and backup on data in the data storage is arranged in the data processing server; the handheld terminal is provided with the wireless communication unit which is contacted with the data storage; and the gateway is respectively and electrically connected with the data processing server and an Ethernet. The electronic work list system and the using method of the electronic work list system have the advantages of avoiding human work errors, being capable of controlling work states in real time and saving resources and being high in automation degree.

The invention discloses an intelligent anti-theft helmet for a shared electric vehicle rider to wear, the helmet comprises a helmet outer shell, a helmet inner shell is arranged in an inner cavity of the helmet outer shell, a cavity is formed between the helmet outer shell and the helmet inner shell, a control device is arranged in the cavity, and the outer surface of the control device is covered with a layer of waterproof cloth; the control device comprises a controller, a pressure sensor and a buzzer alarm, an indicator light is arranged on the outer wall of the helmet shell, and an anti-theft rope of a steel structure is connected to the bottom of the helmet shell; after the shared electric vehicle is powered on, power is supplied to the control device through connection of the power source, the pressure sensor senses whether a rider wears a helmet or not for safe riding and sends the sensing result to the controller, and the controller turns on or off the reminding lamp and the buzzer alarm to remind the rider to wear the helmet. The safety helmet can be prevented from being stolen, and the danger of the shared electric vehicle is reduced.

The invention discloses a continuous learning method, system, equipment and storage medium for image semantic segmentation network. On the one hand, the method of extracting old knowledge representations and aligning them through nonlinear transformation in feature space effectively maintains the invariance of old knowledge and improves the Ability to learn new knowledge. On the other hand, the topology of the new category is optimized in the embedding space, and the invariance of the topology of the old category is maintained to reduce forgetting and prevent confusion between classes. This makes it unnecessary to provide labels of old categories in the continuous learning of semantic segmentation, reducing the cost of labeling. In general, as a universal semantic segmentation continuous learning method, the present invention has no restrictions on application scenarios, and has strong generalization ability and practical value.

The invention relates to a message pushing method and device, electronic equipment and a storage medium, and the method comprises the steps: receiving a to-be-pushed prompt message, recording a receiving moment, obtaining an operation information set of a touch operation inputted by a user before the receiving moment, inputting the operation information set into a preset operation analysis model,and predicting a processing mode of the user for the reminding message by utilizing the operation analysis model, and pushing the reminding message according to a preset pushing strategy correspondingto the processing mode. According to the embodiment of the invention, the processing mode of the user for the reminding message can be predicted according to the touch operation input by the user before the reminding message is received, and the reminding message is pushed according to the predicted processing mode, namely, the processing mode of the user for the reminding message can be predicted according to the state of the user before the user receives the reminding message, and a push strategy conforming to the predicted processing mode is adopted to push the reminding message, so that the push strategy is automatically selected in combination with the user state.

The invention discloses a cardiopulmonary resuscitation instructor and an instruction method based on AR technology. Among them, the cardiopulmonary resuscitation instructor includes: a camera module, used to obtain emergency scene environment information and / or patient information; a virtual image processing module, used to call and / or construct a corresponding virtual first aid guidance video / image; a data analysis and processing module, It is also used to determine the display position of the virtual first aid guidance video / image in the emergency scene environment according to the first aid scene environment information and / or patient information acquired by the camera module; the AR display module is used for displaying and / or in the emergency scene environment The virtual first aid guidance video / image is superimposed on the corresponding position. The present invention can alleviate emergency personnel's forgetting or negligence due to nervousness, and provide first aid guidance step by step; ordinary people who have not been trained can also perform cardiopulmonary resuscitation emergency operations instantly.

The invention discloses an incremental semantic segmentation method for deviation context information correction. The method comprises the following steps: firstly, acquiring semantic segmentation data streams of a plurality of categories, and dividing the semantic segmentation data streams into a plurality of training data sets; in the first incremental semantic segmentation learning step, an initial semantic segmentation network model is obtained through learning of a training data set; in the next incremental semantic segmentation learning step, a new class pixel point erasing method is used for generating a deviation context information corrected picture pair for a newly obtained training data set containing a new class, constructing a deviation context information corrected training data set, and constructing a deviation context information corrected training data set based on the deviation context information corrected training data set. And updating the latest incremental semantic segmentation network model by a learning method of deviation context information correction and adaptive class balance. According to the method, context information, deviating to a new class, of old-class pixel points can be effectively corrected, the class distribution problem of deviation can be relieved, and forgetting of old-class knowledge and semantic drift of background classes are reduced.

The invention discloses a method for realizing NOx emission prediction based on an attention mechanism LSTM (Long Short Term Memory) model. The method comprises the following steps: (1) acquiring data from a power station boiler system operation database; (2) preprocessing the acquired data; (3) carrying out data sorting according to the data format requirement of the LSTM model; (4) using an Attention-LSTM (Long Short Term Memory) model for outputting the NOx emission amount; and (5) decoding the features output by the attention module, and outputting a predicted value. According to the method, the problem that model convergence is difficult due to the fact that the input time period is too long when the LSTM model is established through high-dimensional data can be solved, the phenomenon that important information is forgotten by the model in the modeling process is reduced, and the unit NOx emission prediction precision is improved.

The invention relates to a Bluetooth-based quick verification method and system and equipment. The verification method is based on the verification system. The verification system includes a computerand the mobile terminal equipment. The verification method includes: setting a first password in the computer, generating an encryption / decryption protocol according to equipment information of the mobile terminal equipment, and storing a second password, which corresponds to the first password, to the mobile terminal equipment at the same time; sending broadcast information, which contains the second password, to the computer by the mobile terminal equipment through Bluetooth after the computer identifies Bluetooth information of the mobile terminal equipment; and after the computer obtains the broadcast information and carries out decoding through the encryption / decryption protocol, acquiring an unlocking code corresponding to the first password, and carrying out verification through password management software built in the computer. By adopting the verification method of the invention, verification login can be carried out without the need for recording a large number of passwordsby a user, and a verification process is quicker and more convenient.

The invention belongs to the technical field of intelligent teaching management, and particularly relates to an interactive feedback and teaching integrated management system for real-time online intelligent teaching, which comprises a teaching management system and a classroom interactive feedback system, and is characterized in that the teaching management system comprises a teacher subsystem and a student subsystem; the input end of the classroom interaction feedback system is connected with the output end of the student subsystem, the output end of the classroom interaction feedback system is connected with a personal module and a group module, the output ends of the personal module and the group module are connected with an interaction module, and the interaction module is divided into pre-class interaction, in-class interaction and after-class interaction; the output end of the interaction module is connected with a big data analysis module. According to the interactive feedback and teaching integrated management system for real-time online intelligent teaching, by improving the online rate and learning attention of students, the absorption of the students on teaching knowledge can be further improved, and the learning efficiency of the students is improved.

A storage type clothes hook, the clothes hook is made by bending the plate body, the wide design avoids the shortcomings of traditional clothes hooks from scratching the clothes; the top of the clothes hook is equipped with a tube-shaped boss to prevent the clothes from hanging for a long time Wrinkles are produced; the setting of fluorescent material on the surface of the boss is convenient for users to find the position of the clothes hook when there is no light or the light is very weak, so as to pick and place clothes; The bad habit of carrying important items with you can reduce the trouble of forgetting or missing your belt when you go out.

The invention relates to a voice consultation system and method, the system comprises a video voice system, a health file system, a semantic conversion module, a cloud database and a semantic analysis module, and the method comprises the following steps: a family doctor sends a consultation instruction to the video voice system; the health record system calls health record information of a specified patient; a consultant checks the health record of the specified patient; the voice conversion module acquires voice information of a consultation doctor and a patient according to a preset time range to perform semantic conversion; and the analysis module calls the voice text data of the consultation doctor and the patient stored in the cloud database and performs text extraction according to a preset keyword, and the consultation doctor calls the electronic signature and sends the electronic signature to the health file system. According to the invention, the stored consultation voice information is converted into a text form, so that a doctor can conveniently record the service information of the patient consultation; the problem that a traditional consultation file is easy to lose is solved; and the situation of archive data loss or information forgetting caused by handwritten medical records is reduced.

The invention relates to the technical field of intelligent traffic, in particular to a continuous pedestrian trajectory prediction method based on a memory enhanced network, comprising an auto-encoder for encoding input information, and a controller for reading and writing memory from a key value memory; realizing multi-modal trajectory prediction by using a multi-hop attention mechanism, and generating a prediction trajectory through multi-hop iterative query; a future trajectory is reconstructed through a decoder, a training controller selects some typical key value memories from a current task and stores the typical key value memories in an external memory, and a part of previous data is played back every certain period in the current task. Current and previous key value memories are stored through an external memory, and in order to slow down catastrophic forgetting of a neural network, forgetting of a model to a previous task is slowed down by using a sparse experience playback method; for the uncertainty of future trajectories of pedestrians, a multi-hop attention mechanism is introduced to generate a reasonable multi-modal trajectory, so that the multi-modal trajectory is decoded and output.

The invention relates to a lifetime learning method for solving a catastrophic forgetting problem. The method comprises the steps of (1) obtaining feature information before and after updating of a generation model by means of a coding network of an old solver; (2) whitening the features before updating, and obtaining robust features through linear combination of new and old features for constructing a transformation matrix to whiten the new features so as to obtain orthogonal feature information before and after updating; (3) according to a style migration algorithm, obtaining orthogonal style information before and after updating of the generation model by utilizing the Gram matrix, and updating the generation model by fusing loss items of orthogonal style consistency; and (4) based on aknowledge distillation algorithm, training a solver to be paired with the generated pseudo data to represent an old task, and shuffling the old task with new data for updating the solver. According to the method, the problems that the generation model is forgotten along with increase of tasks in the generation model and classification information is lost due to independent training of a solver can be relieved, and therefore the catastrophic forgetting problem in lifetime learning existing in neural network and artificial intelligence (AI) system development is solved.