154results about How to "Improve training effect" patented technology

The invention provides a scene device for Chinese learning training based on artificial intelligence and virtual reality. The scene device comprises a virtual reality device, an expert system based on the artificial intelligence, a spoken language training module and a Chinese character training module. The virtual reality device comprises a pair of helmet type 3D glasses, a 3D displayer, a virtual electronic pen, a brain wave inductor, a touch screen, a camera and a computer host. According to the scene device, the mood state and the identity information of a trainer are obtained through a face recognition sub-module, whether the trainer focuses attention or not is judged through a brain wave recognition sub-module, guidance training is carried out on the trainer through the cooperation with the artificial intelligence algorithm, and intelligent evaluating and intelligent error correction are particularly carried out through the spoken language training module with the cooperation of the psychological states and the language environment moods of the trainer. Intelligent evaluating and intelligent error correction are carried out through the Chinese character training module with the cooperation of the psychological states and the language environment moods of the trainer, therefore, in the training interactivity is accordingly improved, a feedback type training mode is used in the scene device, and finally the training effect on the trainer is improved.

The invention discloses an anti-cheating method in a face recognition system. The method comprises the following specific steps that: obtaining an image, carrying out normalization processing on the image, and carrying out a feature extraction step, a model design stage, a training stage and a prediction stage, wherein the feature extraction stage is used for extracting seven features including color diversity features, fuzzy degree features, picture moment features, definition features, spectral features, mirror image features and convolution features. By use of the method, a residual-mlp network is cooperated with face micro-textures and a support vector machine, the accuracy and the speed of face living body detection are greatly improved, and in addition, a better detection effect canbe achieved. In addition, by use of the method, hardware except the camera does not need to be added, in addition, super real-time face living body detection speed can be achieved without the operation of personnel to be detected, and the problem in the prior art that living body detection time is long, the hardware needs to be added and detection ability is poor can be solved.

The invention provides a brand name identification method, a computer device and a computer readable storage medium. The brand name identification method comprises the following steps of setting a plurality of labels; obtaining a training text, obtaining a preset brand name from the training text according to the set labeling label, and labeling the preset brand name; obtaining a representation vector of each character in the training text, inputting a character string composed of a plurality of characters into a BERT model, obtaining the representation vector of each character, calculating the probability of each label of each character, using the label with the maximum probability as a target label of the character, and training a brand recognition model; And obtaining a to-be-identifiedtext, inputting the to-be-identified text into the brand identification model, obtaining a representation vector of each character, and calculating a predicted brand name according to the label of each character of the to-be-identified text. The invention also provides a computer device for realizing the brand name identification method and a computer readable storage medium. According to the present invention, the brand name recognition accuracy can be improved, and the manual participation degree is reduced.

The present invention discloses a psychotherapy simulation system based on a virtual reality technology. Through the PC software (psychological training software), a PC is employed to connect with a virtual reality helmet to allow operators to enter the virtual environment and perform psychological disorder training in the virtual environment. Through combination with the biofeedback technology, namely, the help seeker's can be known at will in the treatment process to regulate the stimulating dose on the basis. The psychotherapy simulation system based on the virtual reality technology can allow crowd having psychological illness or obstacle to receive training more easily, can allow the crowd having the psychological illness or obstacle to perform treatment with no need for going to hospital, can allow the crowd having the psychological illness or obstacle to perform training at anytime and anywhere and can allow the crowd having the psychological illness or obstacle to actively perform assessment training.

The invention provides a face skin type analysis method and system based on depth learning and the generation confrontation network. The method comprises steps that the image acquisition step, a to-be-analyzed face image is acquired; the block division step, the to-be-analyzed face image is analyzed, and skin blocks of the to-be-analyzed face image are divided according to organs; the analysis step, different skin analyzers are inputted to different skin blocks and are used for analyzing skin states of the corresponding skin blocks; and the result output step, the analysis results of all the skin analyzers are outputted. The method is advantaged in that on the condition that face skin type data is not easy to acquire and distribution is free, through the generation confrontation network, auxiliary data is generated, a depth learning model is utilized to replace a traditional image processing algorithm for skin type analysis application, face skin details can be analyzed, in combination with the generation confrontation network, for different-characteristic skins, the limited data is utilized to generate training data, and the better training effect is acquired.

The invention discloses a cross-modal processing method and device, electronic equipment and a computer storage medium, and relates to the technical field of natural language processing. According tothe specific implementation scheme, the method comprises the steps that 1, acquiring sample sets, wherein the sample set comprises a plurality of corpora and a plurality of images; generating multipletraining samples, wherein each training sample is a combination of at least one corpus and at least one corresponding image; using multiple training samples, training semantic models, learning the semantic model to obtain a semantic vector of corpus and image combination; adopting a trained semantic model to carry out cross-modal processing between the corpora and the images; and combining the corpora and the corresponding images for training .The semantic model learns the semantic association between the corpus and the corresponding image, the training effect of the semantic model is improved, and the problems that in the prior art, during multi-modal processing, each modal is independently trained, semantic association between different modals is isolated, and the effect of the model obtained through training is poor are solved.

The invention discloses an oil and gas storage and transportation process flow simulating system and method. The and gas storage and transportation process flow simulating system comprises an LNG (Liquefied Natural Gas) receiving station simulating system (73), a multi-purpose station simulating system (74), a natural gas pressurization and transportation simulating system (75), an oil depot simulating system (76), a crude oil long-transportation simulating system (77), a main control machine (78), a graphic machine (79) and a projection system which are connected through a PPI (Processor/Peripheral Interface) bus, wherein the main control machine (78) is connected with the graphic machine (79) through internet, and the PPI bus is connected with the internet through a network interconnecting device. According to the invention, the process flow of oil and gas storage and transportation is visually, vividly and livingly simulated by adopting forms such as sand tables, animations and the like, operations of a student on the sand tables can be dynamically and interactively shown in a flow plate and a projection screen in real time, and the student has a feeling of being personally on the scene, thus training effect is enhanced and training cost is reduced.

The invention relates to an exoskeleton biological feedback hand functional training device which comprises a hand back attaching panel worn on a patient hand, a thumb driving mechanism, four finger driving mechanisms and a motor transmission mechanism. The exoskeleton biological feedback hand functional training device is characterized in that the thumb driving mechanism, the four finger driving mechanisms and the motor transmission mechanism are respectively connected onto the hand back attaching panel, a linear motor transmission arranged in the motor transmission mechanism is connected with the thumb driving mechanism, and a micro direct current motor transmission mechanism is connected with the four finger driving mechanisms. The wearing type design is adopted so that a patient can carry out functional training based on daily life auxiliaries conveniently and the training effect can be improved. An under-actuated multi-joint linked finger mechanism can achieve interphalangeal joint exercise under a light structure. Surface electromyogram signal triggering and dynamic intelligent control achieve interactive biofeedback training. A voice control mode provides passive training to achieve interactive motor imagery therapy. A healthy side hand moves to touch and control hands on both sides to train simultaneously in the active flaccid paralysis and spasm stage.

The invention discloses an upper limb both-arm rehabilitation training human-computer interaction method. The upper limb both-arm rehabilitation training human-computer interaction method comprises atraining scene is selected and displayed; a control command is sent according to the training scene; space positions and rotating angle data of upper limb shoulder joints, elbow joints and wrist joints of a patient under an overall coordination system, and force application data of tail ends of upper limbs are collected in real time; the collected data are resolved and calculated, the upper limb movement trace, the joint angles of shoulders, elbows and wrists, and patient tail end force application data in the upper limb training process of the patient are obtained; and the upper limb movementtrace is displayed in the virtual scene in real time, the patient is guided to conduct track training, and rehabilitation condition of the patient upper limbs is judged through comparison of the training scene and the virtual scene. In addition, the invention further provides an upper limb both-arm rehabilitation training human-computer interaction system, the purpose is that the patient is helped to conduct postoperative upper limb rehabilitation training, targeted training is conducted on the patient conveniently, and the training effect can be improved.

The embodiment of the invention discloses a user classification method and an input method and device. According to the embodiment, a word vector of each word in text data is calculated, a word set corresponding to each target object is determined, and feature vectors finally input into a model are determined according to the word set corresponding to each target object and the word vector corresponding to each word without relying on subjective experience of business personnel. Therefore, the training effect of the model can be improved, and the accuracy of classification or clustering according to the model can also be improved.

The invention discloses an image tampering recognition model training method and device, an image tampering recognition method and an image tampering recognition device, and belongs to the technical field of image recognition, and the method comprises the following steps: obtaining a training image set and a label of each training image in the training image set, when the training image is a tampered image, the label of the training image comprising tampered position information and a tampering type; taking the training image as input, and extracting RGB features, texture features and steganography features of the training image through a feature extraction network layer of a to-be-trained image tampering recognition model; inputting the RGB features, the texture features and the steganography features of the training image into a detection network layer of an image tampering recognition model, and obtaining a tampering recognition result of the training image; and performing difference comparison on the tampering recognition result and the label, optimizing parameters of the image tampering recognition model according to a difference comparison result, continuing training, and ending the training until a predetermined condition is met. According to the embodiment of the invention, the tampered image can be accurately and quickly recognized.

The present invention provides a serum free medium used for cultivating CHO cellular extensively, and discloses the prescription of the culture medium. DMEM/F12 is used as a base culture medium and is added with the components such as vitamin C, transferring, neovaricaine, hydroxyl sodium butyrate and so on. The present invention has the advantages that no animal blood serum is contained and the content of albumen is low; the cultivating effect of the CHO cellular is good, and the CHO cellular can obtain the same effect with the culture medium containing blood serum.

The invention provides an online video intelligent processing system based on deep learning in a cloud environment. The online video intelligent processing system comprises the following components of a video data acquisition layer which acquires video data; a video processing layer which correspondingly performs an offline processing function and an online processing function according to the properties of different video processing tasks; a data service layer which combines an offline processing result and an online processing result for aiming at applications in different fields; and a field service layer which performs visual result feedback to a client. According to the online video intelligent processing system based on deep learning in a cloud environment, video processing, deep learning and cloud calculation are combined; and parallel video processing is realized by means of cloud calculation technology, thereby improving video processing speed. Deep learning is applied in video processing, thereby improving accuracy in video processing.

The invention relates to computer simulation and provides a three-dimension visual simulation method and a three-dimension visual simulation system for driving simulation. The method specifically comprises the following steps: loading a three-dimension scene model for running of a train along a track; refreshing a running image, in which a user drives the train, output to the user at a fixed frequency, wherein each refreshing includes correspondingly updating the three-dimension scene model according to a scene state input signal and correspondingly updating the signal machine state in the three-dimension scene model according to a signal machine state input signal; calculating the viewpoint and the view angle of the user in the three-dimension scene model at the current moment according to a movement state input signal of the train; and refreshing the running image, in which the user drives the train, output to the user according to the updated three-dimension scene model as well as the viewpoint and the view angle. High-fidelity three-dimension visual simulation in different environments is realized, which can help improve the training effect of CBTC training.

The invention belongs to the technical field of sport equipment and particularly relates to a training aiding device for table tennis. The training aiding device for table tennis comprises a table aswell as a serving device, a racket, a monitoring device and a processor which are arranged on one side of the table, wherein the monitoring device comprises an image acquisition unit used for monitoring motion trajectory of a ping-pong ball, a thin film pressure sensor arranged on the racket and a physiological status sensor arranged on the body surface of a player and used for monitoring blood pressure, body temperature and blood oxygen signals of the player. The image acquisition unit, the thin film pressure sensor, the physiological status sensor and the processor are used cooperatively, sothat a training process of the player can be performed again, the player can receive targeted training according to a result output by the training aiding device, and the training effect is improved.

The invention discloses a vital capacity training device for the respiratory medicine department. The vital capacity training device comprises a mask, an exhaust mechanism, a mounting pipe, a slidingmechanism, a monitoring mechanism, an adjusting mechanism, an alarm, a lithium battery and a limiting block. According to the vital capacity training device for the respiratory medicine department, the amount of air exhaled by a patient can be trained by means of the sliding mechanism and the monitoring mechanism matched in use, the amount of air exhaled by the patient can be subjected to enhancedtraining treatment by using the adjusting mechanism of the vital capacity training device, so that the training effect of the postoperative vital capacity of the patient is greatly improved by usingthe vital capacity training device, and the postoperative rehabilitation effect of the patient is improved.

The invention provides a model training method, a text classification method and related devices, and relates to the technical field of machine learning. According to the invention, when the inaccurate supervised learning is carried out on the neural network model by utilizing the training sample with accurate label labeling and the training sample with inaccurate label labeling, a respective corresponding proportion parameter for each training sample is determined; a weight parameter vector representing the accurate labeling probability of all training sample labels is obtained; the feature vectors corresponding to all the training samples are learned by using the weight parameter vectors; proportional output vectors are obtained, a vector based on the proportion is outputted. Compared with the prior art, the model parameters of the neural network model are updated, so that the neural network model can learn more accurate sample information marked by the label when learning is not accurately supervised. The inaccurate sample information marked by the learning label is reduced. The training effect of the neural network model when learning is not accurately supervised is improved.

The invention discloses a teaching, training and examining system based on virtual simulation for the gas industry, which is applied to the gas industry, and a use method of the system. The teaching,training and examining system comprises a teaching block, a practical operation training block and an examining block, wherein the teaching block, the practical operation training block and the examining block comprise a model library, a question library and a theoretical value simulation library, the model library comprises a three-dimensional virtual scene, a virtual tool, a device and a system,a third person figure model and the like, and the teaching block and the practical operation training block are provided with a study reminding module. A network teaching, training and examining platform is constructed according to the virtual simulation system. Due to the arrangement of the system, the flexibility of teaching, practical operation training and examining of the gas industry is effectively improved, the cost is reduced, meanwhile, the training effect is improved, no threat is caused to the personal safety of an operator, the use is safe and reliable, and the extensibility is extremely strong. In addition, for the system, the real-time updating of data can be easily realized, the training content is the latest, and a guarantee is provided for the working competence of new workers.

The invention discloses a vocal-assisted breath practice device comprising a main pipe body and a blow nozzle. The front surface wall of an outer side of the main pipe body is provided with an LED lamp, and a bottom side of the main pipe body is connected to the blow nozzle through a branch pipe. A generator is fixedly connected to the top center of the main pipe body, a bottom of the generator isconnected to an accelerator through a transmission shaft, and a bottom of the accelerator is fixedly connected to a rotating shaft located on a center line in the main pipe body. According to the vocal-assisted breath practice device, a blade can be driven to rotate through airflow, the rotating shaft is driven to rotate through the blade, finally a rotor of the generator rotates and cuts a magnetic induction line to generate an alternating current, the alternating current is connected to an LED lamp, so that the LED lamp flashes and shines, the blowing strength of a user can be intuitively reflected through a flashing speed and brightness of the LED lamp, the enthusiasm of user training can be effectively improved by the structure, the training is intuitive and full of fun, and the training effect is effectively improved.

The invention discloses a method and platform for network training based on virtual imaging and relates to the technical field of methods of augmented reality and network training. The method comprises the steps that imaging scanning and measurement are conducted to sites and facilities involved in training contents; based on field live-thing scanning data, an interactive virtual imaging wandering model is established; a training flow map is made, and rich media courseware and simulation models of each unit in the flow are made; the related rich media courseware and simulation models are implanted into the virtual imaging wandering model, so that an interactive virtual imaging training model is formed; the virtual imaging training model is released to a server; and based on a training flow map, user side software is designed, and the user side software is used to access the interactive virtual imaging training model on the server so as to complete training. According to the scheme, field live-things and training contents are combined organically; training cost is low, contents are visualized, effects are good, and risks are low; and the method and platform are especially suitable for personnel training with complicated processes, many involved sites, environmental risks or confidentiality requirements.

The invention relates to the field of military discipline training or the field of culture amusement, and discloses a shoot combined training system. The shoot combined training system comprises a plurality of training units, wherein the training units carry out data interaction by a wireless communication mode; the training units can be controlled by a preset control mode. The shoot combined training system can be combined by different training units, various real scenes can be simulated, the actual combat feeling of a scene is increased, the capacity of rapidly judging different targets is trained, the strain capacity of a participating trainee for a complex environment and the psychological quality of the participating trainee can be also trained, and the training effect is thus improved.

The invention provides an intelligent strength training device capable of measuring speeds, acceleration, travel and strength. The intelligent strength training device is characterized in that the device is provided with a portal frame, wherein two pulleys are pivoted on the portal frame, a pulley fixing part is arranged between the two pulleys, a middle pulley is pivoted at the pulley fixing part, one end of a weighing sensor is fixedly connected with the pulley fixing part, a screw penetrates through the other end of the weighing sensor and the portal frame to fix the weighing sensor to the portal frame, the middle pulley is provided with a coaxial optical coded disc, and a photosensor corresponding to the optical coded disc is arranged at the pulley fixing part; the weighing sensor and the photosensor are electrically connected with a data acquisition module, and the data acquisition module is electrically connected with a single chip microcomputer; one pull rope winds around the upper sides of the two pulleys and the lower side of the middle pulley, and a balancing weight and an operation part are arranged at two ends of the pull rope respectively. By means of the strength training device, a user can timely get data such as travel of movement actions, movement times, actual strength exerted during movement and the like.

The application provides a training method and device of a speech processing model. The method comprises the following steps: acquiring a training sample and a sample label, wherein the training sample comprises source language speech data, and the sample label comprises a standard target language sentence and a standard source language phonogram sequence; inputting the source language speech datainto an encoder to process, and generating encoding vector; inputting the encoding vector into a first decoder to process, generating a target language sentence, and updating a speech translation model based on the target language sentence and the standard target language sentence; inputting the encoding vector into a second decoder to process, generating a source language phonogram sequence, andupdating an acoustic model based on the source language phonogram sequence and the standard source language phonogram sequence. Through the training method and device of the speech processing moduleprovided by the application, the training effect of the model can be effectively improved.

The invention belongs to the technical field of natural language processing, particularly relates to an end-to-end image text translation method based on multi-task training, and aims to solve the problem that an existing image text translation model is poor in translation performance due to lack of training data and model structure design. The method comprises the following steps: acquiring data to be translated as input data; preprocessing the input data, and after preprocessing, inputting a pre-constructed image text translation model to obtain a translation result corresponding to the input data; wherein the image text translation model comprises a feature extractor and an encoder-decoder. According to the invention, image text translation performance is improved.

The invention relates to a multi-functional motion auxiliary robot for sitting, standing and traveling. The upward and downward reciprocating motion of the saddle and the forward auxiliary motion of the equipment can be realized by driving stretching and closing of front and rear supporting rods through differential speed of front and rear driving wheels, and scientific rehabilitation exercise training, walking motion assistance and protection and other use functions can be provided for the patients with lower limb dysfunction with reference to the space motion state of the hip joint in the healthy human gait process. The left and right symmetrical crankshafts rotate to move the left and right gait levers to realize periodic interlaced swinging back and forth so as to provide leg lifting assistance for the thighs in gait walking. The inertial block unit can provide stable support to the user in the event of accidental unbalanced falling by means of emergency self-locking. The wheelchair function can also be achieved by adjusting the posture from the walking assistance function to the sitting posture function. Operation is easy and multi-functional switching can be realized, the mechanical structure design size is suitable for application in the home environment, and the robot has a broad market application prospect in the field of medical equipment for the aged and disabled.

The invention discloses an annotated data processing method, an annotated data processing device, computer equipment and a computer storage medium, relates to the technical field of data annotation, and aims to adjust the distribution of to-be-annotated data and improve the effect of annotated data on model training. The method comprises the steps of obtaining sample data under each category randomly extracted from machine data to serve as to-be-labeled data; in a process of labeling the to-be-labeled data based on a labeling platform, counting labeled data under each category, and judging whether the labeled data under each category respectively reaches a training standard preset for a classification prediction model or not; if yes, taking the labeled data reaching the training standard category as training data, and inputting the training data into the network model for training to obtain a classification prediction model; and updating the to-be-labeled data according to the prediction probability of the classification prediction model for the test data.

The invention discloses a football training method, and relates to the technical field of football training systems. An internet of things precise positioning system acquires positioning positions andmovement information of footballers and footballs and transmits the positioning positions and the movement information to a cloud server processing platform, and the cloud server processing platforminstantly generates a positioning system position distribution diagram of a football field, the footballers and footballs. Intelligent ball passing guidance software is arranged on the cloud server processing platform, and the footballers wear tactical wristwatches to individually receive a footballer position distribution mini-map, or wear intelligent glasses to receive comprehensive informationand select a video broadcast mode or positioning system footballer distribution map mode according to own requirements. By starting with footballer position judgment, running position control consciousness raising, football field real-time training scientific and technological means upgrading and the like, a God's perspective scientific and technological football training system for internet of things precise positioning, the intelligent glasses, intelligent wearing, AR (augmented reality) and AI (artificial intelligence) computing is built to improve crowd control level of footballer positionand distance macroscopic immersion feeling, overall situation control precision feeling and the like.

The invention relates to a deep binary neural network training method and system. The deep binary neural network training method comprises the steps: initializing a floating point type deep neural network, and obtaining an initialized network model; based on an alternating direction multiplier method, according to the initialized network model, adopting a target propagation algorithm to obtain anoptimized deep neural network with binary activation and floating point type weight; and based on an alternating direction multiplier method, obtaining a deep binary neural network according to the optimized deep neural network. According to the deep binary neural network training method, the framework is optimized through the alternating direction multiplier method; weight and activation are binarized respectively; the coupling effect caused by simultaneous binarization can be reduced; the training effect of the deep binary neural network is improved; and the target propagation algorithm is adopted to optimize the deep neural network with binary activation, so that the problem of difficulty in quantitative optimization of the deep neural network due to infinitesimal quantization process can be reduced.

The embodiment of the invention discloses an orthopedic arm bone recovery exercise device for medical care and particularly relates to the field of medical treatment; and a base is included, a seat cushion portion is arranged in the base, one end of the seat cushion portion is connected with a backrest portion, a groove is formed in the seat cushion portion, an adjusting mechanism is installed inthe groove, a fixed column is connected to positions in the two sides above the seat cushion portion and at the top end of the base, a movable frame is arranged on the front side of the fixed column,lantern rings are connected to positions outside the fixed column and at the two end portions of the movable frame, and a torsion spring sleeves the outer portion of the fixed column between the two lantern rings. According to the orthopedic arm bone recovery exercise device for medical care in the invention, the feet or the legs can be adjusted and supported according to the height of a user or the placement angle of the legs, and multiple groups of parts for recovery exercise of different varicose arms are arranged, so that all the exercise parts of the whole exercise equipment cooperate with one another, and the arms of the user can be fully restored and exercised.

The invention discloses a lightweight excitation model training method based on hierarchical federated learning, and belongs to the technical field of federated learning. The training method comprises the following steps: determining a target end-side user from end-side users according to initial unit data cost information of the end-side users to generate a target end-side user group; calculating an optimal cloud side strategy of the cloud side server according to the initial unit data cost information of each target end side user; calculating an optimal edge side strategy of each edge aggregator according to the optimal cloud side strategy; calculating an optimal end-side strategy of each target end-side user in a target end-side user group corresponding to the current edge aggregator according to the optimal edge-side strategy of each edge aggregator; and training an excitation model according to the optimal cloud side strategy, the optimal edge side strategy and the optimal end side strategy to obtain a trained excitation model. The invention also discloses a training system, an excitation method and an excitation system of the lightweight excitation model based on hierarchical federal learning.