Abstract: A method for training a neural network, includes: training a super network to obtain a network parameter of the super network, wherein each network layer of the super network includes multiple candidate network sub-structures in parallel; for each network layer of the super network, selecting, from the multiple candidate network sub-structures, a candidate network sub-structure to be a target network sub-structure; constructing a sub-network based on target network sub-structures each selected in a respective network layer of the super network; and training the sub-network, by taking the network parameter inherited from the super network as an initial parameter of the sub-network, to obtain a network parameter of the sub-network.

Abstract: A search method for a neural network model structure, includes: generating an initial generation population of network model structure based on multi-objective optimization hyper parameters, as a current generation population of network model structure; performing selection and crossover on the current generation population of network model structure; generating a part of network model structure based on reinforcement learning mutation, and generating a remaining part of network model structure based on random mutation on the selected and crossed network model structure; generating a new population of network model structure based on the part of network model structure generated by reinforcement learning mutation and the remaining part of network model structure generated by random mutation; and searching a next generation population of network model structure based on the current generation population of network model structure and the new population of network model structure.

Abstract: A hypernetwork training method includes: acquiring a multipath neural subnetwork based on a preconstructed initial hypernetwork; training the multipath neural subnetwork to update a weight parameter of each substructure in the multipath neural subnetwork; synchronizing the weight parameter of each substructure in the multipath neural subnetwork to the preconstructed initial hypernetwork; and determining whether the preconstructed initial hypernetwork converges, and if it is determined that the preconstructed initial hypernetwork does not converge, re-executing the acquiring, the training, the synchronizing, and the determining, to obtain a target hypernetwork.

Abstract: Provided are a method and device for data processing and a storage medium. The method includes: inputting labeled training data in a training set into a preset model to be trained and updating the model parameters of the preset model; inputting labeled verification data in a verification set into the preset model after the model parameters are updated to obtain a first prediction label; obtaining a verification loss value based on a difference between the first prediction label and a marked label of the labeled verification data; determining an auxiliary loss value based on current structural parameters of the preset model; determining whether to stop training the preset model based on the verification loss value and the auxiliary loss value; and classifying data to be classified based on a target network model constructed by a network structure included in the trained preset model to obtain a classification result.

Abstract: A method for processing an image includes: an image to be processed with a first resolution is acquired; and the image to be processed is processed by a target neural network model to obtain a target image, the target image being a denoised image with a second resolution, the second resolution being higher than the first resolution, and the target neural network model including a first preset number of convolutional layers and a second preset number of sub-pixel up-sampling portions.

Abstract: The present disclosure relates to an image processing method and apparatus based on a super network, and a computer storage medium. The method can include that a pretrained backbone network is merged with a rear end of a target detection network to obtain a merged super network, the merged super network is trained, Neural Architecture Search (NAS) is performed based on the trained super network to obtain a target detection neural architecture, and an image to be processed is processed by using the target detection neural architecture to obtain an image processing result.

Abstract: A method for processing an image includes: an image to be processed with a first resolution is acquired; and the image to be processed is processed by a target neural network model to obtain a target image, the target image being a denoised image with a second resolution, the second resolution being higher than the first resolution, and the target neural network model including a first preset number of convolutional layers and a second preset number of sub-pixel up-sampling portions.

Abstract: A hypernetwork training method includes: acquiring a multipath neural subnetwork based on a preconstructed initial hypernetwork; training the multipath neural subnetwork to update a weight parameter of each substructure in the multipath neural subnetwork; synchronizing the weight parameter of each substructure in the multipath neural subnetwork to the preconstructed initial hypernetwork; and determining whether the preconstructed initial hypernetwork converges, and if it is determined that the preconstructed initial hypernetwork does not converge, re-executing the acquiring, the training, the synchronizing, and the determining, to obtain a target hypernetwork.

Abstract: A method for training a neural network, includes: training a super network to obtain a network parameter of the super network, wherein each network layer of the super network includes multiple candidate network sub-structures in parallel; for each network layer of the super network, selecting, from the multiple candidate network sub-structures, a candidate network sub-structure to be a target network sub-structure; constructing a sub-network based on target network sub-structures each selected in a respective network layer of the super network; and training the sub-network, by taking the network parameter inherited from the super network as an initial parameter of the sub-network, to obtain a network parameter of the sub-network.

Abstract: A search method for a neural network model structure, includes: generating an initial generation population of network model structure based on multi-objective optimization hyper parameters, as a current generation population of network model structure; performing selection and crossover on the current generation population of network model structure; generating a part of network model structure based on reinforcement learning mutation, and generating a remaining part of network model structure based on random mutation on the selected and crossed network model structure; generating a new population of network model structure based on the part of network model structure generated by reinforcement learning mutation and the remaining part of network model structure generated by random mutation; and searching a next generation population of network model structure based on the current generation population of network model structure and the new population of network model structure.

Abstract: A method for subnetwork sampling is applicable to a hypernetwork topoloyly. The hypernetwork topology includes n layers, each layer includes at least two substructures, and each substructure includes hatch normalization (BN) modules in one-to-one correspondence with the substructures of a closest upper layer, n>0 and n being a positive integer. The method includes: a substructure A(N) of an N-th layer is selected, 1>N?n; a selected substructure A(N-1) of an (N?1)-th layer is determined; a BN module C(B) in one-to-to correspondence with A(N-1) is determined from the substructure A(N); and the substructure A(N) is added into a subnetwork through the BN module C(B).

Abstract: A supernet construction method includes: setting a linear connection unit in at least one layer of a supernet, wherein an input end of the linear connection unit is connected to an upper layer of a home layer of the linear connection unit, and an output end is connected to a lower layer of the home layer of the linear connection unit; an output and an input of the linear connection unit form a linear relationship, where the linear relationship includes a linear relationship other than that the output is equal to the input.

Abstract: A method for processing an image includes: an image to be processed with a first resolution is acquired; and the image to be processed is processed by a target neural network model to obtain a target image, the target image being a denoised image with a second resolution, the second resolution being higher than the first resolution, and the target neural network model including a first preset number of convolutional layers and a second preset number of sub-pixel up-sampling portions.

Abstract: A super network training method includes: performing sub-network sampling on a super network for multiple rounds to obtain a plurality of sub-networks, wherein for any layer of the super network, different sub-structures are selected when sampling different sub-networks, and training the plurality of sub-networks obtained by sampling and updating the super network.