Abstract: A data denoising method and a related device are provided. According to the method, an artificial intelligence technology may be used to perform denoising on data, and any target denoising operation in at least one denoising operation performed on noisy data includes: generating, based on first prediction information and second prediction information, distribution information corresponding to the target denoising operation, where the first prediction information indicates predicted noise between second noisy data and clean data, the second prediction information indicates a square of the predicted noise between the second noisy data and the clean data or indicates a square of a predicted distance between the first prediction information and actual noise, and the actual noise includes actual noise between the second noisy data and the clean data; and sampling denoised data in distribution space to which the distribution information points.

Abstract: A method for visual reasoning. The method includes: providing a network with sets of inputs and sets of outputs, wherein each set of inputs of the sets of inputs mapping to one of a set of outputs corresponding to the set of inputs based on visual information on the set of inputs, and wherein the network comprising a Probabilistic Generative Model (PGM) and a set of modules; determining a posterior distribution over combinations of one or more modules of the set of modules through the PGM, based on the provided sets of inputs and sets of outputs; and applying domain knowledge as one or more posterior regularization constraints on the determined posterior distribution.

Abstract: A method for training neural networks based on energy-based latent variable models (EBLVMs) includes bi-level optimizations based on a score matching objective. The lower-level optimizes a variational posterior distribution of the latent variables to approximate the true posterior distribution of the EBLVM, and the higher-level optimizes the neural network parameters based on a modified SM objective as a function of the variational posterior distribution. The method is used to train neural networks based on EBLVMs with nonstructural assumptions.