309 results about "Training period" patented technology

In described embodiments, adaptive equalization of a signal in, for example, Serializer / De-serializer transceivers by a) monitoring a data eye in a data path with an eye detector for signal amplitude and / or transition; b) setting the equalizer response of at least one equalizer in the signal path while the signal is present for statistical calibration of the data eye; c) monitoring the data eye and setting the equalizer during periods in which received data is allowed to contain errors (such as link initiation and training periods) and periods in which receive data integrity is to be maintained (such as normal data communication).

A system for training an athlete during a training period includes a digital system that is programmed to: receive a plurality of inputs regarding a current state of the athlete and a training goal for the athlete; employ an expert system to generate a training prescription for the athlete; and generate a plurality of control outputs that correspond to the listing of training activities. An exercise apparatus includes at least one activity device that is configured to: facilitate the athlete performing a predetermined exercise; receive the control output from the digital system and to adjust an exercise parameter so as to correspond to the training parameter indicated by the control output; and generate an electronic result output indicative of use by the athlete of the activity device wherein the result output is transmitted to the digital system.

The invention relates to the technical field of artificial intelligence, in particular to training and prediction methods and devices of a federal transfer learning model. The training method comprises the following steps: a first terminal acquires an encrypted migration model issued by a parameter server in an ith training period, wherein the encrypted migration model is generated by the parameter server according to K encrypted sharing models uploaded by K terminals of K participants in an (i-1)th training period; and the first terminal updates an encryption sharing model in a first local neural network model of the first terminal according to the issued encryption migration model in the ith training period, and trains and updates the first local neural network model according to the first data. Therefore, in the federal transfer learning training process, the privacy of the data of each participant is ensured, and the training efficiency of the model is effectively improved, and theuniqueness of each terminal model is improved while the generalization ability of the model is ensured.

A method, system and computer program product detect attempts to send significant amounts of information out via HTTP tunnels to rogue Web servers from within an otherwise firewalled network. A related goal is to help detect spyware programs. Filters, based on the analysis of HTTP traffic over a training period, help detect anomalies in outbound HTTP traffic using metrics such as request regularity, bandwidth usage, inter-request delay time, and transaction size.

An apparatus, method and system are provided for correlated noise reduction, in a trellis decoding environment, such as second generation HDSL, in which crosstalk impairments may be significant. The preferred embodiments provide equalization and correlated noise reduction, utilizing a training period to generate corresponding coefficients, and utilizing two different training error signals. In addition, the apparatus method and system also provide continued and adaptive correlated noise reduction during data transmission, utilizing two additional error signals, a trellis error signal and a tentative error signal. The trellis error signal is a decision error of a selected previous state of a selected trellis path, in which the selected trellis path has a smallest cumulative error of a plurality of trellis paths, and the selected previous state is preferably the immediately previous state. The tentative error signal is formed as a difference between a delayed, tentative symbol decision and a delayed received data signal subsequent to equalization, in which the delay is preferably one symbol time. The various embodiments may be utilized with trellis encoding, with or without data precoding.

The invention discloses a traction substation watchkeeper virtual immersion type training simulation system and method. Virtual three-dimensional scenes and related animations of a traction substation are established with three-dimensional modeling and animation software, the visual and auditory senses of training objects can be effectively enhanced, and the understanding of the training objects on study content can be intensified; the three training patterns of free exercising, situational teaching and simulation checking, and specific task training content are established with a three-dimensional engine and a database, and the training content is high in pertinence and capable of being freely modified; the scene perceptual feelings are further enhanced through virtual reality immersion type helmets and somatosensory interaction equipment. According to the method, the three-dimensional scene of the whole traction substation can be vividly and visually established through three-dimensional visualization simulation, the daily operation of patrolling, switching, overhauling and the like is simulated through service training content based on the three-dimensional scene, safe operation, a high immersion degree, high visualizability and a short training period are achieved in the training process, and good application prospects are achieved.

A reference timing signal apparatus with a phase-locked loop (PLL) has a computer algorithm which adaptively models the multiple frequencies of an oscillator following a training period. The oscillator is part of a PLL and the oscillation frequency thereof is controlled in response to the phase detector output. The computer algorithm processes the control signal applied to the oscillator. The computer algorithm updates the characteristics of the model relating to the aging and temperature of the oscillator, using for example, a Kalman filter as an adaptive filter, in accordance with a cumulative phase error in the PLL calculated during a given time interval. By the algorithm, the subsequent model predicts the future frequency state of the oscillator on which it was trained. The predicted frequency of the model functions as a reference to correct the frequency of the oscillator in the event that no input reference timing signal is available. Also, the calculated phase error is stored and is used while no input reference timing signal or accurate predicted frequency value is available. With the model updating algorithm, oscillators of low stability performance may be used as cellular base station reference oscillator, which is based on satellite systems, for example, GPS, GLONASS or Galileo systems.

A differential serial communication receiver circuit automatically compensates for intrapair skew between received differential signals on a serial differential communication link, with deterministic skew adjustment set during a receiver training period. Intrapair skew refers to the skew within a pair of differential signals, and is hence interchangeable with the term differential skew in the context of this document. During the receiver training period, a training data pattern is received, such as alternating ones and zeros (e.g., a D10.2 pattern as is known in the art), rather than an actual data payload. The differential serial communication receiver circuit includes a differential skew compensation circuit to compensate for intrapair skew. The differential skew compensation circuit receives a pair of complementary differential input signals including a noninverting input signal and an inverting input signal, and in response generates a skew compensated first differential output signal and a skew compensated second differential output signal. The differential skew compensation circuit compares the relative delay of the skew compensated first differential output signal and the skew compensated second differential output signal, and in response delays at least one of the noninverting input signal or the inverting input signal to reduce intrapair skew.

A person monitoring system employs an exemplary method for wirelessly monitoring one or more persons, wherein each person is initially positioned in a respective person support device (PSD) and each PSD is associated with a presence determining device (PDD) operable to detect a presence of a person in the PSD. A portable unit (PU) of the system wirelessly receives status signals during a training period from one or more PDDs. Each status signal includes a respective identifier corresponding to the PDD. The PU stores identifiers included in status signals received during the training period to produce monitored identifiers. After expiration of the training period, the PU wirelessly receives status signals from at least some of the PDDs. The PU then determines whether each monitored identifier is included in at least one respective post-training status signal received during a predetermined time interval and, if not, alerts a user of the PU.

A system and method for efficient data prefetching. A data stream stored in lower-level memory comprises a contiguous block of data used in a computer program. A prefetch unit in a processor detects a data stream by identifying a sequence of storage accesses referencing a contiguous blocks of data in a monotonically increasing or decreasing manner. After a predetermined training period for a given data stream, the prefetch unit prefetches a portion of the given data stream from memory without write permission, in response to an access that does not request write permission. Also, after the training period, the prefetch unit prefetches a portion of the given data stream from lower-level memory with write permission, in response to determining there has been a prior access to the given data stream that requests write permission subsequent to a number of cache misses reaching a predetermined threshold.

A technique to transmit feedback frames from a control point in each slot of an Association-Beamforming Training Period, as specified in a 60 GHz DBand specification, where at least one sector sweep frame is transmitted from a responding station and at least one sector sweep frame is received by the control point, in order to increase the chance of establishing a directional communication link between the control point and the station.

The invention discloses a polyphone model training method for speech synthesis, and a speech synthesis method and device. The method comprises the following steps of processing a voice data set and a text set so as to generate a training corpus set, wherein the text set corresponds to the voice data set, and the training corpus set comprises texts and Pinyin sequences corresponding to the texts; extracting feature information of the texts; and training polyphone models according to the feature information and the Pinyin sequence. According to the polyphone model training method for speech synthesis, in a polyphone model training process, manual labeling on Pinyin of the texts is not required, a training period of the polyphone models is greatly shortened, meanwhile, the circumstance that the trained polyphone models are inaccurate due to wrong manual labeling is avoided, and accuracy of the trained polyphone models is improved.

The invention belongs to the technical field of brain-computer interaction (BCI), and particularly relates to a method and device for classifying active brain-computer interaction system motor imagery tasks. The method and device for classifying active brain-computer interaction system motor imagery tasks resolve the technical problems that an existing classification mode for the active motor imagery tasks is not high in accuracy and long in execution time. The classification method for the active BCI system motor imagery tasks comprises the steps that (1) original brain electrical signals in the training period are collected; (2) processing and characteristic extraction are carried out on the brain electrical signals, and the detection threshold Vd is calculated; (3) collection and state detection are carried out on the brain electric signals in the active imagery period, and the idle state and the imagery motion state of the brain are detected according to the threshold; (4) the motor imagery brain electric characteristics are classified into left hand motor imagery tasks and right hand motor imagery tasks. The brain electric characteristic classification method which combines threshold detection and supporting of a support vector machine is stable and reliable, improves the accuracy of classifying the motor imagery brain electric characteristics, and effectively shortens the classification execution time.

An efficient voice activity detection method and system suitable for real-time operation in low SNR (signal-to-noise) environments corrupted by non-Gaussian non-stationary background noise. The method utilizes rank order statistics to generate a binary voice detection output based on deviations between a short-term energy magnitude signal and a short-term noise reference signal. The method does not require voice-free training periods to track the background noise nor is it susceptible to rapid changes in overall noise level making it very robust. In addition a long-term adaptation mechanism is applied to reject harmonic or tonal interference.

The invention discloses a distributed well drilling simulation system and an operation method. The distributed well drilling simulation system comprises a choke-line manifold (101), a high pressure manifold (102), a blowout preventer console (103), a flow blocking machine console (104), a long-distance console (105), a driller console (106), a teacher console (107) and a figure projecting unit (108), wherein the driller console (106), the long-distance console (105), the blowout preventer console (103), the flow blocking machine console (104), the choke-line manifold (101) and the high pressure manifold (102) are interconnected through a PPI (processor / peripheral interface) protocol, and the teacher console (107) is connected with the PPI protocol through a PPI interface. The operation method comprises the steps of drilling down, pilling out of a hole, drilling, accident and complex condition processing, well shutting, well killing and the like. The invention has the advantages of realizing the simulation of a top driving well drilling simulation with highly simulation, enhancing the liveness of teaching and training, shortening training period and reducing the training cost.

A reference timing signal apparatus with a phase-locked loop has a computer algorithm which adaptively models the multiple frequencies of an oscillator following a training period. The oscillation frequency of the oscillator is controlled in response to a phase detector output. The computer algorithm processes the control signal applied to the oscillator. The computer algorithm updates the characteristics of the model relating to the aging and temperature of the oscillator, using for example, a Kalman filter as an adaptive filter. By the algorithm, the subsequent model predicts the future frequency state of the oscillator on which it was trained. The predicted frequency of the model functions as a reference to correct the frequency of the oscillator in the event that no input reference timing signal is available. In a case of using pre-processing infinite impulse response filters (IIRFs) before the adaptive processor, the time delay caused by the filters are compensated after the adaptive processor. Without pre-processing IIRFs, aging and temperature update rates are adaptively controlled by dynamically changing the rates depending upon the loop condition to achieve a wider tracking bandwidth. With the model updating algorithm, oscillators of low stability performance may be used as cellular base station reference oscillator.

The invention discloses a full-three-dimensional real-time drilling simulation method, comprising the following steps: (1) a three-dimensional drilling solid model bank is built by adopting a three-dimensional animation modeling method; (2) a graphics processor is specially arranged, the graphics processor comprises a graphics drawing program and a scene simulation control program; (3) the scene simulation control program is communicated with an external mast control program according to given data format, so as to obtain the real-time animation commands and data; (4) scene simulation control program sends out an operational instruction to the graphics drawing program, and the graphics drawing program realizes drilling simulation animation drawing and display. In the invention, based on the computer simulation technology, by taking the practical operational flow of the drilling operation filed as reference, the drilling process and the operation method can be simulated vividly, and high-quality graphics animation is generated for the training of technical skills of operating personnel at the drilling site and students, the training effect is improved, the training period is shortened and the training cost is reduced.

A method for transmitting a full-duplex frame includes scheduling a first flexible allocation resource of a frame as a first resource for a second device served by the first device, scheduling a second flexible allocation resource of the frame as a second resource for a third device served by the first device, and generating the frame including the flexible allocation resources and a first half-duplex training period configured to convey a first training signal, where the first half-duplex training period and the first training signal facilitate an estimation of a channel impulse response (CIR) of a communications channel between a transmit antenna of the first device and a receive antenna of the first device. The method also includes transmitting the frame, and simultaneously receiving the frame.

A method for the supervised teaching of a recurrent neutral network (RNN) is disclosed. A typical embodiment of the method utilizes a large (50 units or more), randomly initialized RNN with a globally stable dynamics. During the training period, the output units of this RNN are teacher-forced to follow the desired output signal. During this period, activations from all hidden units are recorded. At the end of the teaching period, these recorded data are used as input for a method which computes new weights of those connections that feed into the output units. The method is distinguished from existing training methods for RNNs through the following characteristics: (1) Only the weights of connections to output units are changed by learning—existing methods for teaching recurrent networks adjust all network weights. (2) The internal dynamics of large networks are used as a “reservoir” of dynamical components which are not changed, but only newly combined by the learning procedure—existing methods use small networks, whose internal dynamics are themselves completely re-shaped through learning.

The invention discloses a model parameter adjustment method and a device. Through applying a technical scheme disclosed by the embodiment of the invention, a risk control decision-making device monitors the performance of a current model, the risk control decision-making device takes model parameters of the current model when the performance reaches an adjustment threshold and acquires preprocessed training data in recent training period, and the risk control decision-making device carries out adjustment on the model parameters of the current model according to the training data and updates the current model according to the adjusted model parameters, so that the risk control decision-making device is enabled to adjust and update the model automatically without changing the model structure and variable on the basis of monitoring for the model performance, the accuracy of the model applied at present is increased, and thus model improvement and optimization are realized with a low cost.

A compressor for waveforms having at least two waveform states separates the waveform samples into waveform state sample vectors for each waveform state. Waveform state encoders encode the waveform state sample vectors separately to provide compressed waveform data. The waveform state encoder selects waveform state pattern vector and associated codes to represent the waveform state sample vectors. The differences between samples of the waveform state sample vector and waveform state pattern vector are calculated and encoded. Encoding can be lossless or lossy. The waveform state pattern vectors and other parameters for compression are determined during a training period. The waveform state encoders detect features in the waveform state sample vectors and waveform state pattern vectors that are useful for common oscilloscope measurements. Typical waveform states include level states and edge states.

System and method embodiments for mobility prediction in a wireless network enable the wireless network to determine the location of a wireless device with minimal transmissions from the wireless device. In an embodiment, the method includes negotiating with a mobile device to determine a mobility prediction algorithm and a condition upon which the mobile wireless device will report the actual location of the mobile device, training the mobility prediction algorithm using prior mobile wireless device location and timestamp information, determining a predicted location of the mobile device using the mobility prediction algorithm, and setting an predicted location for the mobile device at a time as the actual location for the mobile device at the time when failing to receive a location report from the mobile wireless device, wherein the mobile device transmits actual location information after the training period only if the condition is met.

The invention discloses an image coloring method based on a self-attention generative adversarial network. The method comprises the following steps: step 1, training a grayscale image coloring model;2, inputting the gray level images in the training data set into an adversarial network to execute a feature extraction stage, a feature fusion stage, a deconvolution calculation stage and a self-attention learning stage to reconstruct corresponding color images; step 3, comparing the color image reconstructed after the self-attention learning with the corresponding original color image, and calculating a penalty function shown in the specification; and step 4, calculating a penalty function according to the formula shown in the specification, 4, taking the loss function as the optimization loss of the GAN on the basis of the formula shown in the specification, wherein the formula shown in the specification is shown in the specification; and step 5, dividing the training process into a plurality of preset sub-training periods, and sequentially training the sub-training periods by adopting a step-by-step growth strategy to obtain the generator network. According to the method, the colorimage conforming to human subjective visual preferences is reconstructed from a black-white or gray level image by adopting an adversarial generation network, so that the color image is more realistic.

A stress management device is formed in an integrated design that enables pulse measurement, display of results and a power supply gathered into a small finger clip device. The stress management device includes a non-invasive infrared sensor that identifies and measures pulse rate variability and utilizes that information to calculate the power spectrum distribution and create a curve of the results multiple times per second. The device incorporates a display screen on which a graphic image of a pair of lungs filling and emptying to encourage the user into a predetermined breathing pattern known to reduce stress levels during a training period. After the completion of the training period, the device presents a score reflecting the percentage of times the user achieved high, medium and low relaxation states, and a pie chart reflecting the distribution of the states of relaxation calculated during the training period.

The invention discloses a model parameter verification method, device and apparatus in transverse federated learning and a medium, which are used for solving the problems of lower model training efficiency and accuracy caused by incapability of identifying invalid model parameters in the prior art. The method specifically comprises the steps that model parameters reported by all model training apparatuses are issued to other model training apparatuses except the model training apparatuses of the model training apparatuses through a server for error evaluation; the server can obtain an error value set of each model parameter; therefore, according to the error value set of each model parameter; screening invalid model parameters, therefore, in the model training process, when the initial model parameters used in the next model training period are determined, the invalid model parameters can be eliminated, so that the problem that the model training efficiency and accuracy are low due tothe fact that the invalid model parameters cannot be recognized is effectively solved, and the model training efficiency and accuracy are improved.

The invention discloses a smart grid dispatching control system regulation and control integrated join simulation training system. The smart grid dispatching control system regulation and control integrated join simulation training system relates to a simulation training system. For the problems that training of dispatchers and monitors breaks away from actual situations, training quality is not ideal, regulation and control service level cannot be improved effectively, and safe, reliable, economical and efficient operation of a power grid is influenced in the prior art, the invention provides such smart grid dispatching control system regulation and control integrated join simulation training system, which comprises a dispatcher training simulation subsystem for dispatcher training simulation, and a monitor training simulation subsystem for monitor training simulation. The dispatcher training simulation subsystem and the monitor training simulation subsystem respectively comprise a power system simulation module, a regulation and control center simulation module and a teacher system module. The monitor training simulation subsystem shares a part of modules of the dispatcher training simulation subsystem. The technical scheme can improve professional skill training efficiency of the power grid monitors and dispatchers, reduce training period and improve training quality, so that a high-quality staff team can be developed within a short time, and safe, reliable, economical and efficient operation of the power grid is ensured.

The invention discloses a visualized simulation training system and method for an electrified railway traction substation. Three-dimensional simulation modeling software, animation production software and assistant software are utilized to establish a three-dimensional visualized simulation training system, which can effectively activate a plurality of sensory organs of visual sense and auditory sense and the like of training objects to participate in the perception activity, helps to understand and receive the training content, allows students to autonomously select learning content and difficulty level, provides a favorable condition for giving a full play to the initiative and obtaining effective cognition for the training objects, is high in versatility, short in training period, friendly in interactive interface and high in substitution, and has a good application prospect.

The invention relates to the field of artificial intelligence, and discloses a federal learning model training method and device, and the method comprises the steps that a server determines a first terminal and a second terminal according to the connection state of terminals participating in a federal learning model before issuing global parameters of a (k+1)th training period; wherein the first terminal is a terminal which cannot be connected; wherein the second terminal is a normally connected terminal; the server determines second local parameters of M network layers with contribution degrees meeting set conditions in the first terminal; according to the second local parameters of the M network layers of the first terminal in the kth training period and the first local parameters of the N network layers uploaded by the second terminal, the server sends the first local parameters to the kth training period, and determines the global parameters of the N network layers issued to the second terminal in the (k+1)th training period, so as to reduce the influence on the training effect of the model caused by offline of the terminal.

A distributed drilling simulation system includes a choke manifold, a high pressure manifold, a blowout preventer console, a choke console, a remote console, a driller console, a teacher console and a graphic projecting unit, wherein the driller console, the remote console, the blowout preventer console, the choke console, the choke manifold, and the high pressure manifold are interconnected through a PPI (processor / peripheral interface) protocol. The teacher console is connected with the PPI protocol through a PPI interface. A communication program and a main control program run on a main control computer and a graphic processing program runs on a graphic computer. The invention has the advantages of realizing high-degree top driving drilling simulation, enhancing the field sense for teaching and training, shortening the training period and reducing the training cost.

In described embodiments, adaptive equalization of a signal in, for example, Serializer / De-serializer transceivers by a) monitoring a data eye in a data path with an eye detector for signal amplitude and / or transition; b) setting the equalizer response of at least one equalizer in the signal path while the signal is present for statistical calibration of the data eye; c) monitoring the data eye and setting the equalizer during periods in which received data is allowed to contain errors (such as link initiation and training periods) and periods in which receive data integrity is to be maintained (such as normal data communication).