8 results about "Noise rate" patented technology

The application provides a spaceborne array laser radar three-dimensional terrain mapping data comprehensive simulation method and system, and the method comprises the steps of: establishing a spaceborne array laser radar three-dimensional terrain mapping simulation link; constructing a spaceborne array laser radar comprehensive simulation model to generate a spaceborne array laser radar three-dimensional terrain simulation dataset; and adopting noise rate, average signal photon number, detection probability, signal-to-noise ratio and elevation accuracy as multiple evaluation indexes to evaluate the quality of the spaceborne array laser radar three-dimensional terrain mapping simulation data under different parameter configurations. The application designs a spaceborne array laser photon cloud data simulation method for the ground laser angle reflection CCR, generates a set of spaceborne array laser radar three-dimensional terrain simulation dataset taking the parameters of a certain type of domestic satellite as an example, adopts multiple evaluation indexes to evaluate the quality of the spaceborne array laser simulation data under different parameter configurations, and provides a solid foundation for the development of array laser three-dimensional terrain mapping satellites.

The application discloses a laser radar searching system and method. The method is applied to an electronic device, and the method comprises the following steps: a laser radar system comprises a laser emitting device, a laser detecting device, a time-dependent single-photon counting device and the electronic device; the laser detecting device can be adjusted to obtain a time difference between a laser detector searching echo signal and a laser signal emitted by a last laser emitter and a field of view angle; according to the time difference and the field of view angle, the number of photons of the echo signal and a noise rate corresponding to a scene to be searched are determined; according to the number of photons and the noise rate, a current signal-to-noise ratio is determined; when the current signal-to-noise ratio satisfies a signal-to-noise ratio threshold value, a larger searching range for searching a target mobile terminal exists, so that the searching range can be expanded, and the efficiency of searching the target mobile terminal is improved.

The application belongs to the technical field of deep learning, and relates to an audio and video parsing method based on noise label learning, which comprises the following steps: preprocessing original audio and video to obtain a segment-level input sequence; constructing a mutual learning noise-resistant double-flow network; training the mutual learning noise-resistant double-flow network according to a training set; comparing the validation set indicators of two sub-networks in the trained mutual learning noise-resistant double-flow network, and taking the sub-network with the larger validation set indicator as an audio and video parsing model; and parsing through the audio and video parsing model according to a test set to obtain a video prediction result. The mutual learning noise-resistant double-flow network is composed of two sub-networks with the same structure but different initializations, a cross filtering mechanism is executed according to the clean masks generated by the two sub-networks during the training of the mutual learning noise-resistant double-flow network, and the dynamic confidence ratio is gradually reduced through a cosine strategy, so that the problems of high pseudo-label noise rate and easy overfitting noise in the existing audio and video parsing task are solved.

The processor of the control device is configured to perform: correlating the light reception intensity with each detection pixel for the detection distance and acquiring the light reception intensity in each detection cycle (Cd); monitoring a noise rate (R) of a number of noise pixels (Nn) relative to a number of correction pixels in each detection cycle (Cd), the number of correction pixels being a number of correction pixels after subtracting a number of effective pixels (Nv) of detection pixels that detected an effective echo (Erv) determined to be a reflection echo from the object from a total number (ΣN) of detection pixels, the number of noise pixels (Nn) being a number of detection pixels that detected a noise echo (Ern) determined to be a reflection echo from raindrops falling to the outside; and generating detection data (Dd) that informs a rain state of the outside according to the noise rate (R).

The invention discloses an industrial image anomaly detection method, system and device based on dynamic noise estimation, and the method comprises three stages: a first stage, extracting image multi-scale features through a pre-trained convolutional network, screening feature spaces through an LOF algorithm, preliminarily obtaining a high-confidence normal feature set, and obtaining a high-confidence normal feature set; constructing a core feature memory library based on a greedy algorithm; in the second stage, the image anomaly score sequence is analyzed through a PELT change point detection algorithm, the boundary of a normal sample and the boundary of a noise sample are dynamically divided, and the actual label noise rate is estimated; and in the third stage, noise is injected into normal features to generate an adversarial sample, a semi-supervised training strategy is designed to jointly optimize normal and adversarial feature classification boundaries, and an anomaly positioning result is output through a lightweight discrimination network. According to the method, the label noise rate in the training set can be dynamically estimated, the normal samples and the abnormal samples are effectively distinguished, and the robustness of abnormal detection under label noise pollution data is remarkably enhanced.

The invention discloses an intelligent model robustness test method and system based on smooth noise injection, a medium and equipment. The method comprises the following steps: S1, training a classification model based on a noiseless clean data set; s2, performing category prediction on samples in the clean data set by using the trained classification model, and distinguishing simple samples from difficult samples through prediction difficulty based on sample prediction categories and real labels; s3, calculating the number of noise samples according to the number of the simple samples and a set noise rate, comparing the number of the noise samples with the number of the difficult samples, and entering S4 when a first comparison condition is met; s4, samples with the same number as the noise samples are selected from the difficult samples, sample labels of the samples are disturbed, noise samples are generated, and a noise data set is constructed according to the simple samples and the noise samples; and S5, performing an intelligent model robustness test by using the noise data set. According to the invention, the noise data set with any noise rate can be generated based on the noiseless clean data set and tested.

The invention belongs to the technical field of deep learning, and relates to an audio and video analysis method based on noise label learning, which comprises the following steps: preprocessing an original audio and video to obtain a fragment-level input sequence; constructing a mutual learning anti-noise double-current network; training the mutual learning anti-noise double-flow network according to the training set; comparing the verification set indexes of the two sub-networks in the trained mutual learning anti-noise double-current network, and taking the sub-network with the verification set index greater than that of the other sub-network as an audio and video analysis model; and according to the test set, analyzing through the audio and video analysis model to obtain a video prediction result. According to the invention, through a mutual learning anti-noise double-flow network formed by two sub-networks with the same structure and different initializations, a cross filtering mechanism is executed according to clean masks generated by the two sub-networks in the training of the mutual learning anti-noise double-flow network, and a dynamic confidence proportion is gradually reduced through a cosine strategy. The problems that in an existing audio and video analysis task, the false label noise rate is high, and noise overfitting is prone to occurring are solved.

A computing system is provided, including one or more processing devices. The one or more processing devices are configured to receive quantum circuit parameters including a code parameter of an error correction code and a number of T gates included in a quantum circuit. The one or more processing devices are further configured to receive respective decoder parameters of each of a plurality of candidate decoders. The decoder parameters include a physical noise rate of a plurality of physical qubits at which the quantum circuit is configured to be executed and a stopping time of the candidate decoder. The one or more processing devices are further configured to compute respective spacetime costs of the candidate decoders based on the quantum circuit parameters and the decoder parameters. The one or more processing devices are further configured to output a selection of a lowest-spacetime-cost decoder for implementation at a quantum computing device.