A regional precipitation nowcasting system based on cycle-generative adversarial network (GAN) extension includes an input unit configured to receive an input composite hybrid surface rainfall (HSR) image including precipitation information of a region of interest corresponding to a first time, a cycle-GAN configured to generate a resultant composite HSR image including precipitation information of the region of interest corresponding to a second time which comes later than the first time on the basis of the input composite HSR image using a first cycle-GAN and a second cycle-GAN which is complementary to the first cycle-GAN, and an output unit configured to output the resultant composite HSR image as a nowcasting image of the region of interest. The regional precipitation nowcasting system and method based on cycle-GAN extension can ensure robust temporal causality by applying pixel losses to a cycle-GAN.

A method of processing x-ray images comprises training an artificial neural network to process multi-spectral x-ray projections to determine composition information about an object in terms of equivalent thickness of at least one basis material. The method further comprises providing a multi-spectral x-ray projection of an object, wherein the multi-spectral x-ray projection of the object contains energy content information describing the energy content of the multi-spectral x-ray projection, The multi-spectral x-ray projection is then processed with the artificial neural network to determine composition information about the object, and then the composition information about the object is provided

Vehicle perception techniques include obtaining a training dataset represented by N training histograms, in an image feature space, corresponding to N training images, K-means clustering the N training histograms to determine K clusters with respective K respective cluster centers, wherein K and N are integers greater than or equal to one and K is less than or equal to N, comparing the N training histograms to their respective K cluster centers to determine maximum in-class distances for each of K clusters, applying a deep neural network (DNN) to input images of the set of inputs to output detected/classified objects with respective confidence scores, obtaining adjusted confidence scores by adjusting the confidence scores output by the DNN based on distance ratios of (i) minimal distances of input histograms representing the input images to the K cluster centers and (ii) the respective maximum in-class.

A computing system generates a training data set for training the prediction model to detect defects present in a target surface of a target specimen and training the prediction model to detect defects present in the target surface of the target specimen based on the training data set. The computing system generates the training data set by identifying a set of images for training the prediction model, the set of images comprising a first subset of images. A deep learning network generates a second subset of images for subsequent labelling based on the set of images comprising the first subset of images. The deep learning network generates a third subset of images for labelling based on the set of images comprising the first subset of images and the labeled second subset of images. The computing system continues the process until a threshold number of labeled images is generated.

Methods, apparatus, systems and articles of manufacture to improve deepfake detection with explainability are disclosed. An example apparatus includes interface circuitry to receive a media file, machine readable instructions, and at least one processor circuit to be programmed by the instructions to generate, based on a deepfake classification model, a classification score for the media file, obtain a class output from a final convolution feature map of the deepfake classification model, generate an explainability map based on a pooled weighted feature map along a channel dimension of the feature map, and identify the media file as a real or a deepfake media file based on the classification score and the explainability map.

A method of jointly training of a transferable feature extractor network, an ordinal regressor network, and an order classifier network in an ordinal regression unsupervised domain adaption network by providing a source of labeled source images and unlabeled target images; outputting image representations from a transferable feature extractor network by performing a minimax optimization procedure on the source of labeled source images and unlabeled target images; training a domain discriminator network, using the image representations from the transferable feature extractor network, to distinguish between source images and target images; training an ordinal regressor network using a full set of source images from the transferable feature extractor network; and training an order classifier network using a full set of source images from said transferable feature extractor network.

The invention relates to inferring the state of a system of interest having a plurality of indicator values and possibly being heterogeneous in nature. A number of indicator values from a control state and from a comparison state are gathered. From these indicator values, classification power between the control and comparison states (measure of goodness) is computed. Difference values are computed for the indicator values from the system of interest based on the difference to the indicator values from control and comparison states. From a number of these indicators, composite indicators are formed, and composite measures of goodness and composite difference values are computed. A plurality of composite indicators may be formed at different levels. These indicators may be represented as a tree and grouped according to content, and at the same time they may be arranged according to the measure of goodness or some other value. The indicators, measures of goodness, and difference values may be visualized and shown to a user, who may use such a representation for inferring the state of the system.

A product description generating system dynamically generates a description of an item based on one or more images. The product description generating system includes an attribute machine learning system that receives images of the item and determines at least one attribute value and corresponding prediction confidence value based on the image data. The user review interface provides an interface for reviewing attribute values generated by the attribute machine learning system. If the prediction confidence value is outside a threshold, then the attribute value is reviewed by a user via the user review interface. If, on the other hand, the prediction confidence value is within a threshold, then the attribute value is used to generate the description without user review. The user review interface groups items to be reviewed by taxon and facilitates user review of many items at scale.

A method for detecting a product for defects implemented in an electronic device includes detecting images of a product for defects by a first defect detection model in a preset period, and obtaining a detection result; when a ratio of the number of negative sample images is greater than a preset threshold, training an autoencoder model; obtaining historical positive sample images of the product, inputting the history positive sample images into the trained autoencoder model, and calculating a latent feature; inputting the latent feature of each history positive sample image into a decoding layer of the trained autoencoder model, and calculating newly added positive sample images; training the first defect detection model and obtain a second defect detection model; and inputting images of a product to be detected to the second defect detection model, and obtaining a detection result of the product.

Multi-label classification is improved by determining thresholds and/or scale factors. Selecting thresholds for multi-label classification includes sorting a set of label scores associated with a first label to create an ordered list. Precision and recall values are calculated corresponding to a set of candidate thresholds from score values. The threshold is selected from the candidate thresholds for the first label based on target precision values or recall values. A scale factor is also selected for an activation function for multi-label classification where a metric of scores within a range is calculated. The scale factor is adjusted when the metric of scores are not within the range.