447 results about "State sequence" patented technology

Procedures are described for enhancing target system execution integrity determined by power fingerprinting (PFP): by integrating PFP into the detection phase of comprehensive defense-in-depth security; by deploying a network of PFP enabled nodes executing untrusted devices with predefined inputs forcing a specific state sequence and specific software execution; by embedding module identification information into synchronization signaling; by combining signals from different board elements; by using malware signatures to enhance PFP performance; by automatic characterization and signature extraction; by providing secure signature updates; by protecting against side-channel attacks; performing real-time integrity assessment in embedded platform by monitoring their dynamic power consumption and comparing it against signatures from trusted code, including pre-characterizing power consumption of the platform by concentrating on trace sections carrying the most information about the internal execution status; by using PFP from sequence of bit transitions to detect deviations from authorized execution of software in a digital processor.

The invention relates to a heterogeneous multi-agent collaborative decision-making method based on a depth deterministic policy gradient, belonging to the collaborative decision-making field of a heterogeneous intelligent unmanned system, comprising the following steps of: firstly, defining heterogeneous multi-agent characteristic attributes and reward and punishment rules, defining multi-agent state space and action space, and constructing multi-agent motion environment for collaboratively making decision; then, establishing an actor module for decision-making action and a critic module for evaluating feedback based on the depth-deterministic strategy gradient algorithm, and training the parameters of the learning model; using the trained model to obtain the multi-agent state sequence; and evaluating the situation of the multi-agent motion state sequence according to the reward and punishment rules set in the environment. The invention may construct reasonable sports environment according to actual needs, achieve the purpose of intelligent sensing and strategy optimization through the synergy between multiple agents in the system, and has a positive effect on the development of the unmanned system field in China.

A method of speech driven lip synthesis which applies viseme based training models to units of visual speech. The audio data is grouped into a smaller number of visually distinct visemes rather than the larger number of phonemes. These visemes then form the basis for a Hidden Markov Model (HMM) state sequence or the output nodes of a neural network. During the training phase, audio and visual features are extracted from input speech, which is then aligned according to the apparent viseme sequence with the corresponding audio features being used to calculate the HMM state output probabilities or the output of the neutral network. During the synthesis phase, the acoustic input is aligned with the most likely viseme HMM sequence (in the case of an HMM based model) or with the nodes of the network (in the case of a neural network based system), which is then used for animation.

A re-configurable test circuit for use in an automated test equipment includes a test circuit, a test processor and a programmable logic device. The pin electronics circuit is configured to interface the re-configurable test circuit with a DUT. The test processor includes a timing circuit configured to provide one or more adjustable-timing signals having adjustable timing. The programmable logic device is configured to implement a state machine, a state sequence of which depends on one or more input signals received from the pin electronics circuit, to provide an output signal, which depends on a current or previous state of the state machine, to the pin electronics circuit in response to the signal(s) received from the pin electronics circuit. The test processor is coupled to the programmable logic device to provide at least one of the adjustable-timing signal(s) to the programmable logic device to define timing of the programmable logic device.

The invention relates to the field of the intelligent transportation system, and especially relates to short-term traffic flow forecasting with use of a parameter value sequence of a road segment. A short-term traffic flow forecasting method based on a hidden Markov model comprises the following steps: collected data are processed and counted; a prediction window is set; a starting time measured value of the prediction window, and an average parameter value and a sequence contrast ratio in the prediction window are discretized, so as to form a hidden state and an observation state set of the hidden Markov model; and a Baum Welch algorithm combined with training data are used for learning the model parameter. Finally, for a certain prediction window, a Viterbi algorithm is used to obtain an optimal hidden state sequence based on a known observation state sequence, and a last state of the optimal hidden state sequence is a prediction state. The short-term traffic flow forecasting method based on the hidden Markov model can be used to predict the short-term traffic state in the future, and is an effective method of prediction of the short-term traffic state.

An improved motor controller for driving polyphase alternating current dynamoelectric machines from a direct current power source that employs two inverter bridges controlled with a state vector modulation symmetrical state sequence that has a desired “chop” frequency so that each of their outputs produce a fundamental component that is exactly in phase but chop frequency components that are exactly 180 degrees out of phase with each other, with their outputs combined through respective interphase transformers to average the potentials of the chop frequency components.

The invention provides a server-side simulating system based on an IEC61850 and a method thereof, which are applied to digital substations adopting the IEC61850 for communication and traditional substations. The system consists of a simulator terminal and a simulator console. The method comprises the following steps: loading an SCD model at the simulator terminal for analyzing; breaking out a CID model file needed by an IED; carrying out man-machine interaction in a tree diagram and tabular form to allocate one data generator for each module data; generating state sequence through the data generator; downloading the state sequence to the simulator control table; loading the CID model broken out by the simulator terminal on the simulator console for analyzing; establishing an IEC61850 database, organizing the trackback path information of the IEC61850 model orderly; addressing the IEC61850 database according to the model trackback path corresponding the data generator; refreshing the IEC61850 database state in real time; and interactively communicating with a station monitor and control layer through the IEC61850, thereby finishing the simulation of the IED server terminal. The invention can carry out simulating preview on the whole substation IED, greatly saves project debugging time and has the characteristics of strong operability, high efficiency, flexible configuration and the like.

A single combination task lamp and flashlight, providing separate flood and spot light beams, independently controlled in a three-state sequence by simple push button switches. The two kinds of light beams are produced by separate arrays of compact light emitting devices. both arrays are driven by a single, rechargeable battery powered electrical circuit that provides separate, regulated constant currents to the respective arrays of LEDs. the optics and electronics are constructed in a single, ruggedized, compact module. The module is enclosed within a slim, rugged housing and easily field replaceable with minimal tools.

The invention provides a speech recognition model establishing method based on bottleneck characteristics and a multi-scale and multi-headed attention mechanism, and belongs to the field of model establishing methods. A traditional attention model has the problems of poor recognition performance and simplex attention scale. According to the speech recognition model establishing method based on thebottleneck characteristics and the multi-scale and multi-headed attention mechanism, the bottleneck characteristics are extracted through a deep belief network to serve as a front end, the robustnessof a model can be improved, a multi-scale and multi-headed attention model constituted by convolution kernels of different scales is adopted as a rear end, model establishing is conducted on speech elements at the levels of phoneme, syllable, word and the like, and recurrent neural network hidden layer state sequences and output sequences are calculated one by one; and elements of the positions where the output sequences are located are calculated through decoding networks corresponding to attention networks of all heads, and finally all the output sequences are integrated into a new output sequence. The recognition effect of a speech recognition system can be improved.

Embodiments of the invention relate to methods for generating a multilingual acoustic model. A main acoustic model comprising a main acoustic model having probability distribution functions and a probabilistic state sequence model including first states is provided to a processor. At least one second acoustic model including probability distribution functions and a probabilistic state sequence model including states is also provided to the processor. The processor replaces each of the probability distribution functions of the at least one second acoustic model by one of the probability distribution functions and / or each of the states of the probabilistic state sequence model of the at least one second acoustic model with the state of the probabilistic state sequence model of the main acoustic model based on a criteria set to obtain at least one modified second acoustic model. The criteria set may be a distance measurement. The processor then combines the main acoustic model and the at least one modified second acoustic model to obtain the multilingual acoustic model.

A method and apparatus are disclosed for improving the processing time of reduced complexity sequence estimation techniques, such as reduced state sequence estimation. Precomputing the branch metrics for all possible symbol combinations in the channel memory makes it possible to remove the branch metrics unit and decision-feedback unit from the feedback loop, thereby reducing the critical path. A set of multiplexers select the appropriate branch metrics based on the survivor symbols in the corresponding survivor path cells. The computational load of the precomputations is reduced for multi-dimensional trellis codes by precomputing each dimension of the multi-dimensional trellis code separately. A hybrid survivor memory architecture is also disclosed for a RSSE for a channel having a channel memory of length L, where the survivors corresponding to the L past decoding cycles are stored in a register exchange architecture (REA), and survivors corresponding to later decoding cycles are stored in a trace-back architecture (TBA) or REA.

The embodiment of the invention provides an image rendering processing method and device. The method comprises: performing state detection of a target head, and generating a target state sequence; when it is determined that the target head enters a movement state, simulating the target state sequence, and generating a fitting curve; determining the angle of a visual field of a target scene according to the frame delay time and the fitting curve generated in advance; and performing rendering of the target scene based on the angle of the visual field, and generating the rendering image. According to the embodiment of the invention, the image rendering processing method and device are able to predicate the movement state of the target head through a fitting curve and compensate the estimated deviation of the angle of the visual field. Therefore the deviation of the angle of the visual field may be effectively reduced when the rendering of the image frame starts and ends, the user's spinning sensation when the user's head is rapidly moved may be effectively mitigated, and the good image display effect is obtained.

The invention discloses a sequence recommendation method based on adaptive attention. The method comprises the following steps: determining input adaptive attention at an input layer of an adaptive attention perception GRU network obtained by pre-training; applying the input adaptive attention to the history interactive item sequence to obtain the input sequence; Based on the combination of historical interactive item sequence and input sequence, the output result of input layer is obtained. Hidden adaptive attention is determined in that hidden state layer, and the hidden adaptive attention is acted on the output result of the input layer to obtain a hidden state sequence; the output result of the hidden state layer is obtained; At the output layer of the adaptive attention-aware GRU network, the recommended items are determined according to the output results of the hidden state layer. According to the technical proposal provided by the embodiment of the invention, the recommendationperformance is improved. The invention also discloses a sequence recommendation device based on adaptive attention, which has corresponding technical effects.

The present video tracking technique outputs a Maximum A Posterior (MAP) solution for a target object based on two object templates obtained from a start and an end keyframe of a whole state sequence. The technique first minimizes the whole state space of the sequence by generating a sparse set of local two-dimensional modes in each frame of the sequence. The two-dimensional modes are converted into three-dimensional points within a three-dimensional volume. The three-dimensional points are clustered using a spectral clustering technique where each cluster corresponds to a possible trajectory segment of the target object. If there is occlusion in the sequence, occlusion segments are generated so that an optimal trajectory of the target object can be obtained.

The invention discloses an early-stage drowning behavior detection method based on videos, which includes the following steps: 1 defining swimming characteristic parameters and obtaining a plurality of state parameters for describing swimmer behaviors with a data integration method; 2 obtaining video image sequence of normal swimming behaviors and imitating swimming drowning behaviors in advance, respectively collecting swimmers and respectively establishing and calculating hidden Markov models of swimmers to obtain drowning behavior state sequence and normal behavior state sequence; 3 acquiring video image sequence in real time and collecting swimmers; and 4 establishing and calculating the hidden Markov model of each swimmer to obtain current behavior state sequence, and judging whether a swimmer drowning behavior exists or not according to current behavior state sequence, drowning behavior state sequence and normal behavior state sequence. If the swimmer drowning behavior exists, an alarm is given out. The early-stage drowning behavior detection method based on videos has the advantages of being accurate and reliable in detection, good in robustness, high in noise immunity and good in adaptability to illumination changes.

A method and system are directed to enabling a detector to perform relatively accurate channel estimates on blocks of data that include relatively short training sequences in wireless communication signals. Channel re-estimation is performed by transferring the estimation calculations from the time domain into a channel state domain. In one embodiment, information obtained from known training sequences and unknown data may be combined to generate an initial channel estimate. An embodiment of a new recursive Expectation Maximization (EM) process generally i) initializes the EM process with the initial channel estimate; ii) determines updates for the channel estimates; iii) produces a reduced observation vector; iv) generates a special sequence of channel states; v) rearranges each channel state estimate in the reduced observation vector; vi) generates a low-dimensional channel model matrix inversion based on the generated special sequence of channel states; and vii) generates channel estimates based on the generated matrix inversion.

A single combination task lamp and flashlight instrument, providing separate flood and spot light beams, independently controlled in a three-state sequence by simple push button switches. The two kinds of light beams are produced by separate arrays of compact light emitting devices. both arrays are driven by a single, rechargeable battery powered electrical circuit that provides separate, regulated constant currents to the respective arrays of LEDs. The optics and electronics are constructed in a single, ruggedized, compact module. The module is enclosed within an elongated handle that positions the center of mass for optimum balance of the instrument and the control switches for convenient operation according to the sense of touch.

The invention discloses a method for making temperature early warning for a wind generating set based on BootStrap confidence coefficient calculation and a system of the method, and belongs to the field of wind generating sets. The method includes the following steps that real-time data is preprocessed, and working condition recognition is performed on a real-time data variable and multiple divided sub working conditions in a preset database; under the corresponding sub working conditions, real-time temperature data is subjected to data state marking through a designated threshold value limit,abnormal point marks and normal marks are included, and a state sequence in a time sequence is formed; for the state marking sequence formed in the time sequence, the number Nn of the abnormal pointmarks of N continuous data is counted, and an abnormity rate eta is calculated; the triggering condition of confidence coefficient calculation is a preset abnormity rate limit value etaa, and when theabnormity rate eta exceeds the abnormity rate limit value etaa, BootStrap confidence coefficient calculation is triggered, so that the level of a confidence coefficient is calculated; when the levelof the confidence coefficient exceeds an early warning index CLlimit, failure early warning is performed. Calculation of the abnormity rate and calculation of the level of confidence are integrated, so that the rate of a false alarm is greatly decreased.

A joint detection-decoding receiver using a multistage parallel interference canceller for selectively estimating a partial codeword corresponding to a state transition (branch) is disclosed. The JDD receiver of the DS-CDMA system includes a PIC that employs a state sequence to selectively carry out a symbol estimation of a partial codeword corresponding to the branch so as to reduce complexity and a calculation load as well as maintaining a high performance of the receiver according to the joint detection-decoding.

A reduced state sequence estimator, and associated method, for equalizing data symbols communicated during operation of a communication system. The estimator is formed of multi-dimensional states of groups of symbols that are combined over more than one dimension to the multi-dimensional. Once the groups of symbols are formed, the groups are partitioned into partition sets. Equalization operations are performed over multiple time epochs of the trellis formed of the partitioned groups of symbols over multiple time epochs.

The invention relates to a risk control method for deducing an illegal behavior of medical insurance through data analysis, and the method comprises the steps: 1, obtaining the current and historicalclinical consultation behavior data of a risk control object and the personal information and public data correlated with the risk control object; 2, obtaining the features correlated with the clinical consultation behavior of the risk control object from the data obtained at step 1; 3, classifying the clinical consultation behavior of each clinical consultation in the current and historical clinical consultation behaviors of the risk control object into a corresponding clinical consultation class according to the extracted features, and forming a clinical consultation class sequence; 4, taking the clinical consultation class sequence formed at step 3 as an observation sequence, taking a clinical consultation purpose as a hidden state, calculating the most possible hidden state sequence according to a hidden Markov model, wherein the most possible hidden state sequence comprises the most possible clinical consultation purpose of the risk control object; 5, outputting the medical data correlated with the current and historical clinical consultation behaviors of the risk control object if the most possible clinical consultation purpose, in the hidden state sequence, of the risk control object is corresponding to an abnormal purpose.

The invention belongs to the field of wireless communication, and discloses a key generation method capable of ensuring the security of wireless communication. The key generation method is based on a wireless channel reciprocity and a curve fitting method, extracts a shared key of wireless communication users from a shared wireless channel state to ensure the communication security, and specifically comprises five steps consisting of channel state estimation, pretreatment, quantization and coding, information coordination and privacy amplification; after the treatment of privacy amplification, a generated bit string can be used as a final shared key for transmitting and receiving parties, for protection the subsequent communication security. According to the method, the problem that a difference exists between channel state sequences collected by the transmitting and receiving parties can be solved, key negotiation and whole key generation efficiency are ensured, and the method can be more widely applied to various complex actual application scenarios, for efficiently ensuring the communication security.

Portions of an input measurement sequence are classified into a plurality of regimes by associating each of a plurality of dynamic models with one a switching state such that a model is selected when its associated switching state is true. In a Viterbi-based method, a state transition record is determined, based on the input sequence. A switching state sequence is determined by backtracking through the state transition record. Finally, portions of the input sequence are classified into different regimes, responsive to the switching state sequence. In a variational-based method, the switching state at a particular instance is also determined by a switching model. The dynamic model is then decoupled from the switching model. Parameters of the decoupled dynamic model are determined responsive to a switching state probability estimate. A state of the decoupled dynamic model corresponding to a measurement at the particular instance is estimated, responsive to the input sequence. Parameters of the decoupled switching model are then determined responsive to the dynamic state estimate. A probability is estimated for each possible switching state of the decoupled switching model. A switching state sequence is determined based on the estimated switching state probabilities. Finally, portions of the input sequence are classified into different regimes, responsive to the determined switching state sequence.

A method and apparatus are provided for using the uncertainty of a noise-removal process during pattern recognition. In particular, noise is removed from a representation of a portion of a noisy signal to produce a representation of a cleaned signal. In the meantime, an uncertainty associated with the noise removal is computed and is used with the representation of the cleaned signal to modify a probability for a phonetic state in the recognition system. In particular embodiments, the uncertainty is used to modify a probability distribution, by increasing the variance in each Gaussian distribution by the amount equal to the estimated variance of the cleaned signal, which is used in decoding the phonetic state sequence in a pattern recognition task.