Abstract: A method is provided for data processing performed by a processing system. The method comprises determining a set of first tokens for first data and a set of second token for second data, each token comprising information associated with a segment of the respective data, determining pair-wise similarities between the set of first tokens and the set of second tokens, each pair comprising a first token in the set of first tokens and a second token in the set of second tokens, determining, for each first token in the set of first tokens, a maximum similarity based on the determined pair-wise similarities between the respective first token and the second tokens in the set of second tokens, and determining a first similarity between the first data and the second data by aggregating the maximum similarities corresponding to the first tokens in the set of first set of tokens.

Abstract: A neural network training method includes performing, in a forward propagation process, binarization processing on a target weight by using a binarization function, and using data obtained through the binarization processing as a weight of a first neural network layer in a neural network; and calculating, in a backward propagation process, a gradient of a loss function with respect to the target weight by using a gradient of a fitting function as a gradient of the binarization function.

Abstract: Embodiments of this disclosure relate to the field of artificial intelligence, and disclose a feature extraction method and apparatus. The method includes: obtaining a to-be-processed object, and obtaining a segmented object based on the to-be-processed object, where the segmented object includes some elements in the to-be-processed object, a first vector indicates the segmented object, and a second vector indicates some elements in the segmented object; performing feature extraction on the first vector to obtain a first feature, and performing feature extraction on the second vector to obtain a second feature; fusing at least two second features based on a first target weight, to obtain a first fused feature; and performing fusion processing on the first feature and the first fused feature to obtain a second fused feature, where the second fused feature is used to obtain a feature of the to-be-processed object.

Abstract: Embodiments of this application disclose a convolutional neural network pruning processing method, a data processing method, and a device, which may be applied to the field of artificial intelligence. The convolutional neural network pruning processing method includes: performing sparse training on a convolutional neural network by using a constructed objective loss function, where the objective loss function may include three sub-loss functions.

Abstract: An image processing method is disclosed in embodiments of this disclosure and is applied to the field of artificial intelligence. The method includes: obtaining an input feature map of an image to be processed, where the input feature map includes a first input sub-feature map and a second input sub-feature map, and resolution of the first input sub-feature map is higher than resolution of the second input sub-feature map; performing feature fusion processing on the input feature map by using a target network, to obtain an output feature map, where a feature of the first input sub-feature map is fused to a feature of the second input sub-feature map from a low level to a high level in the target network; and performing, based on the output feature map, object detection on the image to be processed, to obtain an object detection result.

Abstract: This disclosure discloses a perception network. The perception network may be applied to the artificial intelligence field, and includes a feature extraction network. A first block in the feature extraction network is configured to perform convolution processing on input data, to obtain M target feature maps; at least one second block in the feature extraction network is configured to perform convolution processing on M1 target feature maps in the M target feature maps, to obtain M1 first feature maps; a target operation in the feature extraction network is used to process M2 target feature maps in the M target feature maps, to obtain M2 second feature maps; and a concatenation operation in the feature extraction network is used to concatenate the M1 first feature maps and the M2 second feature maps, to obtain a concatenated feature map.

Abstract: Embodiments of this application disclose a model structure, a method for training a model, an image enhancement method, and a device, and may be applied to the computer vision field in the artificial intelligence field. The model structure includes: a selection module, a plurality of first neural network layers, a segmentation module, a transformer module, a recombination module, and a plurality of second neural network layers. The model overcomes a limitation that the transformer module can only be used to process a natural language task, and may be applied to a low-level vision task. The model includes the plurality of first/second neural network layers, and different first/second neural network layers correspond to different image enhancement tasks. Therefore, after being trained, the model can be used to process different image enhancement tasks.

Abstract: A neural network training method, includes: obtaining an input feature map of a training image; performing feature extraction processing on the input feature map by using a feature extraction core of a neural network to obtain a first candidate feature map; adding the first candidate feature map and a second candidate feature map to obtain an output feature map, where the second candidate feature map is a feature map obtained after a value corresponding to each element in the input feature map is increased by N times, and N is greater than 0; determining an image processing result of the training image based on the output feature map; and adjusting a parameter of the neural network based on the image processing result.

Abstract: The technology of this application relates to a neural network distillation method, applied to the field of artificial intelligence, and includes processing to-be-processed data by using a first neural network and a second neural network to obtain a first target output and a second target output, where the first target output is obtained by performing kernel function-based transformation on an output of the first neural network layer, and the second target output is obtained by performing kernel function-based transformation on an output of the second neural network layer. The method further includes performing knowledge distillation on the first neural network based on a target loss constructed by using the first target output and the second target output.

Abstract: A data feature extraction method and apparatus in the field of artificial intelligence are provided. An addition convolution operation is performed to extract a target feature in quantized data based on quantized feature extraction parameters, that is, to calculate a sum of absolute values of differences between the quantized feature extraction parameters and the quantized data, to obtain the target feature based on the sum. In addition, feature extraction parameters and data are quantized by using a same quantization parameter. According to this application, a storage resource is saved, a computing resource is saved, thereby reducing a limitation on an application of artificial intelligence to a resource-limited device. Further, when the extracted feature data is dequantized, the feature data may be dequantized based on the quantization parameters.

Abstract: A neural network training method in the artificial intelligence field includes: inputting training data into a neural network; determining a first input space of a second target layer in the neural network based on a first output space of a first target layer in the neural network; and inputting a feature vector in the first input space into the second target layer, where a capability of fitting random noise by the neural network when the feature vector in the first input space is input into the second target layer is lower than a capability of fitting the random noise by using an output space that is in the neural network and that exists when a feature vector in the first output space is input into the second target layer. This application helps avoid an overfitting phenomenon that occurs when the neural network processes an image, text, or speech.

Abstract: This application provides a neural network model compression method in the field of artificial intelligence. The method includes: obtaining, by a server, a first neural network model and training data of the first neural network that are uploaded by user equipment; obtaining a PU classifier based on the training data of the first neural network and unlabeled data stored in the server; selecting, by using the PU classifier, extended data from the unlabeled data stored in the server, where the extended data has a property and distribution similar to a property and distribution of the training data of the first neural network model; and training a second neural network model by using a knowledge distillation (KD) method based on the extended data, where the first neural network model is used as a teacher network model and the second neural network model is used as a student network model.

Abstract: This application relates to an image recognition technology in the field of computer vision in the field of artificial intelligence, and provides an image classification method and apparatus. The method includes: obtaining an input feature map of a to-be-processed image; performing convolution processing on the input feature map based on M convolution kernels of a neural network, to obtain a candidate output feature map of M channels, where M is a positive integer; performing matrix transformation on the M channels of the candidate output feature map based on N matrices, to obtain an output feature map of N channels, where a quantity of channels of each of the N matrices is less than M, N is greater than M, and N is a positive integer; and classify the to-be-processed image based on the output feature map, to obtain a classification result of the to-be-processed image.

Abstract: This application relates to an image recognition technology in the field of computer vision of artificial intelligence, and provides an image classification method and apparatus. The method includes: obtaining an input feature map of a to-be-processed image; performing feature extraction processing on the input feature map based on a feature extraction kernel of a neural network, to obtain an output feature map, where each of a plurality of output sub-feature maps is determined based on the corresponding input sub-feature map and the feature extraction kernel, at least one of the output sub-feature maps is determined based on a target matrix obtained after an absolute value is taken, and a difference between the target matrix and the input sub-feature map corresponding to the target matrix is the feature extraction kernel; and classifying the to-be-processed image based on the output feature map, to obtain a classification result of the to-be-processed image.

Abstract: The present application discloses an image generation method, a neural network compression method, and a related apparatus and device in the field of artificial intelligence. The image generation method includes: inputting a first matrix into an initial image generator to obtain a generated image; inputting the generated image into a preset discriminator to obtain a determining result, where the preset discriminator is obtained through training based on a real image and a category corresponding to the real image; updating the initial image generator based on the determining result to obtain a target image generator; and further inputting a second matrix into the target image generator to obtain a sample image. Further, a neural network compression method is disclosed, to compress the preset discriminator based on the sample image obtained by using the foregoing image generation method.

Abstract: The present application discloses a neural network search method in the field of artificial intelligence, and the neural network search method includes: obtaining a feature tensor of each of a plurality of neural networks, where the feature tensor of each neural network is used to represent a computing capability of the neural network; inputting the feature tensor of each of the plurality of neural networks into an accuracy prediction model for calculation, to obtain accuracy of each neural network, where the accuracy prediction model is obtained through training based on a ranking-based loss function; and determining a neural network corresponding to the maximum accuracy as a target neural network. Embodiments of the present invention help improve accuracy of a network structure found through search.

Abstract: This application discloses a convolutional neural network-based image processing method and image processing apparatus in the artificial intelligence field. The method may include: receiving an input image; preprocessing the input image to obtain preprocessed image information; and performing convolution on the image information using a convolutional neural network, and outputting a convolution result. In embodiments of this application, the image processing apparatus may store primary convolution kernels of convolution layers, and before performing convolution using the convolution layers, generate secondary convolution kernels using the primary convolution kernels of the convolution layers.

Abstract: This application discloses a convolutional neural network-based image processing method and image processing apparatus in the artificial intelligence field. The method may include: receiving an input image; preprocessing the input image to obtain preprocessed image information; and performing convolution on the image information using a convolutional neural network, and outputting a convolution result. In embodiments of this application, the image processing apparatus may store primary convolution kernels of convolution layers, and before performing convolution using the convolution layers, generate secondary convolution kernels using the primary convolution kernels of the convolution layers.

Abstract: The present invention provides a method and an apparatus for determining an identity identifier of a face in a face image, and a terminal. The method includes: obtaining an original feature vector of a face image; selecting k candidate vectors from a face image database; selecting a matching vector of the original feature vector from the k candidate vectors; and determining, an identity identifier that is of the matching vector. In embodiments of the present invention, a face image database stores a medium-level feature vector formed by means of mutual interaction between a low-level face feature vector and autocorrelation and cross-correlation submatrices in a joint Bayesian probability matrix. The medium-level feature vector includes information about mutual interaction between the face feature vector and the autocorrelation and cross-correlation submatrices in the joint Bayesian probability matrix, so that efficiency and accuracy of facial recognition can be improved.

Abstract: An image processing method and apparatus are disclosed. The method includes obtaining a two-dimensional target face image, receiving an identification curve marked by a user in the target face image, locating a facial contour curve of a face from the target face image according to the identification curve and by using an image segmentation technology, determining a three-dimensional posture and a feature point position of the face in the target face image, and constructing a three-dimensional shape of the face in the target face image according to the facial contour curve, the three-dimensional posture, and the feature point position of the face in the target face image by using a preset empirical model of a three-dimensional face shape and a target function matching the empirical model of the three-dimensional face shape.