1303 results about "Adversarial network" patented technology

The invention discloses a single image super-resolution reconstruction method based on a conditional generative adversarial network. A judgment condition, namely an original real image, is added intoa judger network of the generative adversarial network. A deep residual error learning module is added into a generator network to realize learning of high-frequency information and alleviate the problem of gradient disappearance. The single low-resolution image is input to be reconstructed into a pre-trained conditional generative adversarial network, and super-resolution reconstruction is performed to obtain a reconstructed high-resolution image; learning steps of the conditional generative adversarial network model include: learning a model of the conditional adversarial network; inputtingthe high-resolution training set and the low-resolution training set into a conditional generative adversarial network model, using pre-trained model parameters as initialization parameters of the training, judging the convergence condition of the whole network through a loss function, obtaining a finally trained conditional generative adversarial network model when the loss function is converged,and storing the model parameters.

A method and apparatus for automated liver segmentation in a 3D medical image of a patient is disclosed. A 3D medical image, such as a 3D computed tomography (CT) volume, of a patient is received. The 3D medical image of the patient is input to a trained deep image-to-image network. The trained deep image-to-image network is trained in an adversarial network together with a discriminative network that distinguishes between predicted liver segmentation masks generated by the deep image-to-image network from input training volumes and ground truth liver segmentation masks. A liver segmentation mask defining a segmented liver region in the 3D medical image of the patient is generated using the trained deep image-to-image network.

The invention discloses a zero sample image classification method based on combination of a variational autocoder and an adversarial network. Samples of a known category are input during model training; category mapping of samples of a training set serves as a condition for guidance; the network is subjected to back propagation of optimization parameters through five loss functions of reconstruction loss, generation loss, discrimination loss, divergence loss and classification loss; pseudo-samples of a corresponding unknown category are generated through guidance of category mapping of the unknown category; and a pseudo-sample training classifier is used for testing on the samples of the unknown category. The high-quality samples beneficial to image classification are generated through theguidance of the category mapping, so that the problem of lack of the training samples of the unknown category in a zero sample scene is solved; and zero sample learning is converted into supervised learning in traditional machine learning, so that the classification accuracy of traditional zero sample learning is improved, the classification accuracy is obviously improved in generalized zero sample learning, and an idea for efficiently generating the samples to improve the classification accuracy is provided for the zero sample learning.

A system is provided for unsupervised cross-domain image generation relative to a first and second image domain that each include real images. A first generator generates synthetic images similar to real images in the second domain while including a semantic content of real images in the first domain. A second generator generates synthetic images similar to real images in the first domain while including a semantic content of real images in the second domain. A first discriminator discriminates real images in the first domain against synthetic images generated by the second generator. A second discriminator discriminates real images in the second domain against synthetic images generated by the first generator. The discriminators and generators are deep neural networks and respectively form a generative network and a discriminative network in a cyclic GAN framework configured to increase an error rate of the discriminative network to improve synthetic image quality.

Methods and apparatus for automated medical image analysis using deep learning networks are disclosed. In a method of automatically performing a medical image analysis task on a medical image of a patient, a medical image of a patient is received. The medical image is input to a trained deep neural network. An output model that provides a result of a target medical image analysis task on the input medical image is automatically estimated using the trained deep neural network. The trained deep neural network is trained in one of a discriminative adversarial network or a deep image-to-image dual inverse network.

The invention discloses a face image enhancement method based on a generative adversarial network (GAN). The method comprises the following steps of: 1, preprocessing face images having multiple posesby using a 3D dense face alignment method; 2, designing a face enhancement network based on the GAN, and generating the GAN in two steps; 3, according to a task requirement, designing an objective function 4 corresponding to the Step-I and the Step-II, pre-training a recognition model by using MS-1-celeb, and pre-training a TS-GAN model by using augmented data; and 5, using Multi-PIE as a training Set, using a back propagation algorithm (4) to complete the pre-trained TS-GAN model parameters until convergence. The finally trained TS-GAN model can obtain a front face image corresponding to aninput image. Further, the front face image retains original illumination, is true in visual degree, and retains original identity information.

The invention discloses a semantic image restoration method based on a DenseNet generative adversarial network. The method comprises the following steps of preprocessing collected images, and constructing a training and testing data set; constructing a DenseNet generative adversarial network; training a DenseNet generation adversarial network stage; and finally, realizing repair processing of thedefect image by using the trained network. Under the framework of the generative adversarial network, the DenseNet structure is introduced, a new loss function is constructed to optimize the network,the gradient disappearance is reduced, network parameters are reduced, meanwhile, transmission and utilization of features are improved, and the similarity and visual effect of large-area semantic information missing image restoration are improved and improved. Examples show that face image restoration with serious defect information can be achieved, and compared with other existing methods, the restoration results better conform to visual cognition.

The invention discloses an image classification method based on a confrontation network generated through feature recalibration. The image classification method based on the confrontation network generated through feature recalibration is suitable for the field of machine learning and comprises the steps that to-be-classified image data are input into a confrontation network model for network training; a generator and a discriminator which are constituted by a convolutional network are constructed; random noise is initialized and input into the generator; the random noise is subjected to multilevel deconvolution operation in the generator through the convolutional network, and finally, generated samples are obtained; the generated samples and authentic samples are input into the discriminator; and the input samples are subjected to convolution and pooling operation in the discriminator through the convolutional network, thus a feature graph is obtained, a compressed and activated SENetmodule is imported into an intermediate layer of the convolutional network to calibrate the feature graph, thus the calibrated feature graph is obtained, global average pooling is used, and finally,image data classification is output. The SENet module is imported into the intermediate layer of the discriminator, the importance degree of each feature channel is automatically learned, useful features relevant to a task are extracted, features irrelevant to the task are restrained, and thus semi-supervised learning performance is improved.

The invention discloses a multi-task named entity recognition and confrontation training method for medical field. The method includes the following steps of (1) collecting and processing data sets, so that each row is composed of a word and a label; (2) using a convolutional neural network to encode the information at the word character level, obtaining character vectors, and then stitching withword vectors to form input feature vectors; (3) constructing a sharing layer, and using a bidirection long-short-term memory nerve network to conduct modeling on input feature vectors of each word ina sentence to learn the common features of each task; (4) constructing a task layer, and conducting model on the input feature vectors and the output information in (3) through a bidirection long-short-term network to learn private features of each task; (5) using conditional random fields to decode labels of the outputs of (3) and (4); (6) using the information of the sharing layer to train a confrontation network to reduce the private features mixed into the sharing layer. According to the method, multi-task learning is performed on the data sets of multiple disease domains, confrontation training is introduced to make the features of the sharing layer and task layer more independent, and the task of training multiple named entity recognition simultaneously in a specific domain is accomplished quickly and efficiently.

The present invention requires to protect a conditional generative adversarial network-based online handwriting identification method. The method comprises the steps of 101 using a user registration module to register the basic information of a user; 102 using a reception module to receive a section of character information inputted by the user, wherein the information comprises the character writing style, the character writing strength and the character writing spacing; 103 training a conditional generative adversarial network on a handwriting signature data set by taking the category labels as the conditions, and being able to generate the corresponding directional digital features according to the information of the category labels; 104 using a handwriting identification module, using the conditional generative adversarial network to mine the personalized handwriting of the user and using an adversarial network signature discrimination model D which is a dichotomy device to discriminate whether the inputted data is the real handwriting data or a generated sample; 105 using an application module to apply the handwriting identification to an access control system and a plurality of user document signing scenes. The conditional generative adversarial network-based online handwriting identification method of the present invention has higher stability, safety and convenience, at the same time, can identify the handwriting style, strength and spacing information of the users by combining a conditional generative adversarial network method, and avoids the problem that the character features are not extracted completely.

The invention discloses a face image repairing method based on a generative adversarial network. The method comprises the following steps that: (1) searching a great quantity of images which contain complete and clear faces, and establishing a face image database; (2) constructing the generative adversarial network; (3) training the generative adversarial network, and optimizing the parameters ofa generator and a descriminator in the generative adversarial network; and (4) inputting random vectors which obey normal distribution into the trained generator, generating the face image, comparingthe intact area of the face image to be repaired with the corresponding area of the generated image, continuously regulating the input vectors until the intact area of the face image to be repaired and the corresponding area of the generated image are similar, and finally, replacing the pixel value of a blocked or damaged area in the face image to be repaired by the pixel value of the corresponding area for generating the face image. By use of the method, the generative adversarial network with a deep learning structure is adopted by aiming at the problem that the blocked or damaged face imageis repaired, and therefore, an image repairing problem in image processing is effectively solved.

The invention discloses a super-resolution reconstruction method based on a conditional generative adversarial network, and the method specifically comprises the steps: making a low-resolution image and a corresponding high-resolution image training set by using a disclosed super-resolution image data set; constructing a conditional generative adversarial network model, using dense residual blocksin the generator network, and realizing super-resolution image reconstruction at the tail end of the generation network model by using a sub-pixel up-sampling method; inputting the training image setinto a conditional generative adversarial network for model training, and enabling a training model to converge through a perception loss function; carrying out down-sampling processing on the imagetest set to obtain a low-resolution test image; and inputting the low-resolution test image into the conditional adversarial network model to obtain a high-quality high-resolution image. The method can well solve the problems that a super-resolution image generated by a traditional generative adversarial network looks like clear, and evaluation indexes are extremely low, and meanwhile, the problems of gradient disappearance and high-frequency information loss are relieved through a dense residual network.

The invention discloses a pedestrian re-identification method based on a generative adversarial network image super-resolution technology. The method includes the following steps: utilizing a Laplace pyramid generative adversarial network to generate a group of clear images, utilizing local maximum event expression and a Dense Correspondences algorithm to separately extract HSV color features, texture features and LAB color features of the images, carrying out fusion on the features, utilizing a cross vision quadratic discriminant analysis algorithm to conduct metric learning on the features, utilizing a Manhattan distance to calculate the distance between a probe set and a gallery set, and finally utilizing a multi-shot mode to carry out 1:N and N:N evaluation. A LAPGAN network is utilized to generate high-resolution image, and then a conventional method is utilized to obtain the features of the images and carry out corresponding matching. Deeping learning and the conventional method are combined to solve the problem of low image resolution caused by illumination, angle and other reasons, and an image matching rate is increased.

The invention discloses a face image aging synthesis method based on a feature separation adversarial network, belongs to the technical field of computer vision, and relates to a face image aging synthesis method. The method comprises the steps: supposing that the age features of a face are distributed in a Manifold, extracting the hidden features of an original image through an Auto-encoder, adding a corresponding age condition, finally generating an adversarial network through the condition, synthesizing a face image of a specific age, and maintaining the identity corresponding to an original face image. The innovativeness of the invention lies in that the method achieves the constraint of the hidden features outputted by an Encoder through the intra-class distance measure, enables the hidden features to be exclusively correlated with the identity, and removes the information correlated with the age, so as to guarantee that the age of the generated image is matched with a target agein a better way after the age is added. The method can be used for the face image aging synthesis, the cross-age face recognition and beatifying software.

The invention belongs to the field of video and image processing. The objective of the invention is to further improve the reconstruction effect and reconstruction precision of a high-resolution image, promote the structure of the generative adversarial network and the improvement of a loss function; the invention discloses a dense connection generative adversarial network single image super-resolution reconstruction method. A generation network and an adversarial network are included. A basic framework of a residual dense network RDN is adopted by the generation network; the adversarial network adopts a deep convolution generative adversarial network DCGAN discriminator network framework; the low-resolution image is used as an input and is sent into a generation network for processing; and the obtained output is sent to the adversarial network for judgment, a judgment result is fed back to the generative network through a loss function, the steps are repeated until the adversarial network is judged to be qualified, the generative network can generate a clear image, and then super-resolution reconstruction of a low-resolution image is completed by using the trained generative network. The method is mainly applied to image processing occasions.

Systems and methods for detecting patterns in data from a time-series and for detecting outliers in network data in an unsupervised manner are provided. In one implementation, a method includes the steps of obtaining network data from a network to be monitored and creating a window from the obtained network data. The method also includes the step of detecting outliers of the obtained data with respect to the window using an unsupervised deep learning process (e.g., using a Generalized Adversarial Network (GAN) learning technique and / or a Bidirectional GAN (BiGAN) learning technique) for enabling the learning of a data distribution. The unsupervised process, for example, does not require manual intervention.

The invention provides a large-amplitude face straightening method by means of an adversarial network and a three-dimensional morphological model. The main content of the large-amplitude face straightening method by means of an adversarial network and a three-dimensional morphological model includes a reconstruction module, a generation network and classification module, and an identification module. The large-amplitude face straightening method by means of an adversarial network and a three-dimensional morphological model includes the steps: a generator generates a forward front face image by taking a non-forward front face image as input, and at the same time a classifier tries to determine whether the image is a real image and utilizes the fed back information to promote the image generated from the generator to be more close to the real image, and at the same time an identification engine is used to maintain the original identity characteristics in the input image. The large-amplitude face straightening method by means of an adversarial network and a three-dimensional morphological model can process the non forward face, especially a large-amplitude deflected face image, can provide a generation network and a morphological model to straighten the face, and can greatly improve the effect of face identification and straightening at the same time.

Described herein are systems and methods for a general, scalable, end-to-end framework that uses a generative adversarial network (GAN) objective to enable robust speech recognition. Encoders trained with the proposed approach enjoy improved invariance by learning to map noisy audio to the same embedding space as that of clean audio. Embodiments of a Wasserstein GAN framework increase the robustness of seq-to-seq models in a scalable, end-to-end fashion. In one or more embodiments, an encoder component is treated as the generator of GAN and is trained to produce indistinguishable embeddings between labeled and unlabeled audio samples. This new robust training approach can learn to induce robustness without alignment or complicated inference pipeline and even where augmentation of audio data is not possible.

The invention discloses an abnormity detection method for satellite telemetry data through an adversarial network autoencoder, and the method comprises the steps: breaking the limitation of a traditional empirical model, and employing a pure data driving model; on the basis of a variational autoencoder, introducing a confrontation network idea, using a bidirectional LSTM (Long Short Term Memory) (Long-short term memory network) as a discriminator, and judging whether satellite telemetry data is abnormal or not by using errors of reconstructed data and original data; aiming at the redundancy problem of a satellite sensor, the conventional situation is broken through, and a Markov distance is used for measuring a reconstruction error. In combination with periodicity of satellite orbit operation, a dynamic threshold determination method based on a periodic time window is provided. The method has the advantages that pure data driving is adopted, expert experience is not needed, and the method can be suitable for various occasions; By combining the respective advantages of the variational auto-encoder and the generative adversarial network, the proposed network has the characteristics of high training speed and relatively easy convergence; eliminating redundant data influence between satellite telemetry data by adopting a Mahalanobis distance. According to the periodicity of the satellite, the dynamic threshold method based on the periodic time window is provided, and the misjudgment rate is reduced.

The invention discloses a short-term traffic flow prediction method based on a generative confrontation network. The prediction method can be applied in an urban traffic road network, wherein historical traffic data is modeled into a matrix sequence, a depth model is used for learning spatial-temporal correlativity in the data, a generator and discriminator in the network are iteratively trained,the trained model can predict the traffic conditions of all road segments in the road network in a short time in the future. The short-term traffic flow prediction method based on the generative confrontation network fully utilizes the spatial-temporal characteristics of the traffic data, the predicted object is extended from a single road segment to the entire traffic network, and the accuracy ofthe prediction is significantly improved.

The invention discloses an image denoising method based on a generative adversarial network, and the method is characterized by comprising the following steps: 1, selecting an experiment data set; 2,selecting Gaussian additive white noise as a noise model; 3, building a generative network model, and training a generator network G for denoising; 4, establishing a discrimination network model, wherein a discriminator D is used for carrying out authenticity classification on the input image; 5, constructing a joint loss function model; 6, training a generative adversarial network; and step 7, carrying out image denoising quality evaluation. According to the image denoising method based on the generative adversarial network, the denoising effect of reserving more texture details and edge features can be achieved.

The invention discloses a method for realizing facial expression migration of a cyclic generative adversarial network based on spectrum normalization, which comprises the following steps of: S1, collecting various facial expression pictures, and classifying the facial expression pictures one by one according to facial expressions; S2, preprocessing the pictures, removing blurred pictures, obtaining five key points of the human face by using a human face detection algorithm, and uniformly cutting face pictures according to the key points; S3, constructing a cyclic generative adversarial networkconsisting of a generator and a discriminator, respectively inputting the two types of preprocessed pictures into the network to calculate a loss function, and training the loss function; and S4, obtaining the trained generator as a tool for human face expression migration, and applying the trained generator to actual measurement. The cyclic generative adversarial network based on spectrum normalization can enable one generator to realize migration of multiple facial expressions, and the generated facial expressions can be more natural and have better robustness.

The invention relates to an adversarial network based rapid multi-visual-angle 3D synthesis-display method and device. A depth map and an RGB picture in certain visual angle are processed, a generative adversarial network (GAN) is constructed to obtain continuous multi-visual-angle 2D mapping of an object, and reconstruction of a 3D model is not needed. Only the depth map and RGB picture in certain visual angle are processed, continuous multi-visual-angle 2D mapping of the object can be obtained needless of reconstructing the high-quality 3D model, the visual angle transformed to obtain the 2Dimage in real time, the flow of 3D object display is simplified, and the 3D object can be displayed in real time in multiple visual angles in higher quality.

The invention discloses a generative and adversarial network-based polarimetric SAR image classification method, and mainly solves the problem of low classification precision caused by unreasonable selection of polarimetric SAR image features in the prior art. Unlabeled and labeled samples are fully utilized. A complex cost function does not need to be designed. The method comprises the followingimplementation steps of (1) inputting a coherence matrix of to-be-classified polarimetric SAR images; (2) performing filtration; (3) generating a sample set; (4) selecting samples; (5) constructing generative and adversarial networks; (6) training the generative and adversarial networks; (7) classifying the sample set; and (8) outputting category labels of all the samples in the sample set of thepolarimetric SAR images. The method has the advantage of remarkable polarimetric SAR image classification effect, and can be further used for target detection and target identification of the polarimetric SAR images.

One embodiment facilitates generating synthetic data objects using a semi-supervised GAN. During operation, a generator module synthesizes a data object derived from a noise vector and an attribute label. The system passes, to an unsupervised discriminator module, the data object and a set of training objects which are obtained from a training data set. The unsupervised discriminator module calculates: a value indicating a probability that the data object is real; and a latent feature representation of the data object. The system passes the latent feature representation and the attribute label to a supervised discriminator module. The supervised discriminator module calculates a value indicating a probability that the attribute label given the data object is real. The system performs the aforementioned steps iteratively until the generator module produces data objects with a given attribute label which the unsupervised and supervised discriminator modules can no longer identify as fake.

The disclosure relates to an image completion method and device. The method includes the following steps: generating n layers of Gauss pyramids on the basis of an image to be restored, n being a positive integer; through a block matching method, completing images of each of the n layers of Gauss pyramids, and obtaining an actual image of each layer after completion; generating a generative network of an adversarial network on the basis of the n image of the n layers of Gauss pyramids and the actual images of all the layers; and on the basis of the n image and the generative network, completing the image to be restored, and obtaining a completed image after completion. According to the technical scheme, the generative network is obtained through a Gauss pyramid of the image to be restored, the actual images of all layers of Gauss pyramids and the adversarial network, so a clear and complete image high in resolution can be generated quickly. Therefore, the resolution of the completed image is improved, learning about the generative network is accelerated, and a more real image can be obtained.

The invention discloses a face restoration method based on a generative adversarial network. The method comprises the following steps: S1, collecting face data and preprocessing the face data; s2, establishing an adversarial network model, wherein the adversarial network model comprises two deep neural networks: a generation network G and a discrimination network D, and generating a face image through the generation network G; judging whether the image is true or false through the judgment network D; and S3, performing face restoration: randomly adding a mask to the test image, simulating a real image defect area, inputting the defect image into a generation network G to generate a face image, replacing the mask area of the generated image with a corresponding position of a missing image,and performing Poisson fusion. Parameters of the network are iteratively updated by utilizing context loss and context loss and global and local discrimination loss, and more natural and realistic face images can be generated and complemented.

The invention relates to a bearing fault diagnosis method based on a semi-supervised generative adversarial network, and the method comprises the following steps of obtaining vibration signals of thebearing in different states, and dividing the vibration signals into multiple samples; randomly dividing the samples into a training set and a test set; constructing a small number of label samples ofdifferent faults in the training set; constructing a one-dimensional semi-supervised generative adversarial network model; and inputting the training set into the adversarial network for training, wherein the trained adversarial network is used to test the diagnosis of centralized bearing faults. The method provided by the invention directly inputs the originally collected vibration signal and directly outputs the category of the bearing fault in the test set through training to achieve an end-to-end optimal diagnosis model, and uses a one-dimensional convolutional layer and a one-dimensionaldeconvolutional layer to enhance the ability of the one-dimensional semi-supervised generative adversarial network for extracting features. The invention is a semi-supervised training method, which does not require a large number of manual label samples, greatly saves time and labor costs, and has strong bearing fault diagnosis effect and anti-noise capability, and good stability.