Abstract: Methods and systems for advanced and augmented training of deep neural networks (DNNs) using synthetic data and innovative generative networks. A method includes training a DNN using synthetic data, training a plurality of DNNs using context data, associating features of the DNNs trained using context data with features of the DNN trained with synthetic data, and generating an augmented DNN using the associated features.

Abstract: A technique for performing data parallel training of a neural network model is disclosed that incorporates batch normalization techniques using partial populations to generate normalization parameters. The technique involves processing, by each processor of a plurality of processors in parallel, a first portion of a sub-batch of training samples allocated to the processor to generate activations for the first portion of the sub-batch. Each processor analyzes the activations and transmits statistical measures for the first portion to an additional processor that reduces the statistical measures from multiple processors to generate normalization parameters for a partial population of the training samples that includes the first portion from each of the plurality of processors. The normalization parameters are then transmitted back to each of the processors to normalize the activations for both the first portion and a second portion of the sub-batch of training samples allocated to each processor.

Abstract: The present disclosure relates to training a machine learning model to classify images. An example method generally includes receiving a training data set including images in a first category and images in a second category. A convolutional neural network (CNN) is trained using the training data set, and a feature map is generated from layers of the CNN based on features of images in the training data set. A first area in the feature map including images in the first category and a second area in the feature map where images in the first category overlap with images in the second category are identified. The first category is split into a first subcategory corresponding to the first area and a second subcategory corresponding to the second area. The CNN is retrained based on the images in the first subcategory, images in the second subcategory, and images in the second category.

Abstract: Provided are a method for transmitting a vehicle-to-everything (V2X) message by a terminal in a wireless communication system, and a device using the method. The method comprises: performing an operation of excluding candidate resources on the basis of a PSSCH-RSRP threshold value in a selection window; determining whether the ratio of remaining candidate resources in the selection window is smaller than a predetermined threshold value; when the ratio is smaller than the threshold value, according to whether at least one of the delay requirement and priority of the V2X message satisfies a predetermined condition, increasing the PSSCH-RSRP threshold value and performing an operation of excluding candidate resources on the basis of the increased PSSCHRSRP threshold value; and transmitting the V2X message by using a resource selected among the remaining candidate resources in the selection window.

Abstract: A method for unlocking an electronic device includes unlocking a screen of the electronic device and displaying a first interface in response to receiving an image, wherein the image comprises an eyeprint image or an iris image that matches a preset condition, wherein the first interface comprises a previous interface that displayed before the screen was last locked, and unlocking the screen and displaying a second interface in response to receiving the image and in response to detecting an action of blinking or moving an eyeball, wherein the second interface is different than the first interface.

Abstract: According to one embodiment, medical signal processing apparatus includes a processing circuit. The processing circuit adjust a level of activation of a unit included in a learned model in accordance with classification of an imaging condition for a process target medical signal. The processing circuit generates an output signal by applying the learned model in which the level of activation has been adjusted, to the medical signal.

Abstract: The present invention provides for a system and method for the diagnosis of skin cancer, wherein visual data is acquired from a skin lesion by a passive or active optical, electronical, thermal or mechanical method, processing the acquired visual field image into a classifier, applying a dermoscopic classification analysis to the image and converting, by sonification techniques, the data to raw audio signals, wherein the raw audio signals are analyzed with a second machine learning algorithm to increase interpretability and the precision of diagnosis.

Abstract: A system and method for detecting human presence in or absence from a field-of-view of a camera by analyzing camera data using a processor inside of or adjacent to the camera itself. In an example, the camera can be integrated with or embedded in another edge-based sensor device. In an example, a video signal processing system receives image data from one or more image sensors and uses a local processing circuit to process the image data and determine if a human being is or is not present during a particular time, interval, or sequence of frames. In an example, the human being identification technique can be used in security or surveillance applications such as for home, business, or other monitoring cameras.

Abstract: A mobile station may comprise a receiver configured to receive, from a base station, a broadcast signal indicating a first subframe configuration for use in an uplink direction and a downlink direction and a second subframe configuration for use in at least a downlink direction. The mobile station may be configured to receive first downlink data and transmit first uplink data, during a single subframe of one or more radio frames, using the first subframe configuration. The mobile station may receive second downlink data, during a single subframe of the one or more radio frames, using the second subframe configuration. The first subframe configuration and the second subframe configuration may be different OFDM subframe configurations and a first subframe configured in accordance with the first subframe configuration and a second subframe configured in accordance with the second subframe configuration may have a same duration.

Abstract: A method and apparatus for monitoring formation of ice on a wind turbine blade are provided. The method includes: capturing an image of the blade through a camera; detecting a region of the blade from the captured image; clearing image information of a background region from the captured image, which is in the captured image except for the region of the blade, to obtain a blade image; and inputting the obtained blade image into a recognition model of ice on a blade obtained by training on a sample set, to determine whether ice is on the captured blade, wherein the sample set comprises a plurality of blade images indicating that ice is on blades.

Abstract: A method for generating training data includes: defining a corresponding category tag for an object model in a three-dimensional scenario; acquiring a plurality of corresponding scenario images by modifying scenario parameters of the three-dimensional scenario; rendering the object model to a monochromic material corresponding to the category tag of the object model; acquiring a semantic segmentation image corresponding to each scenario image based on the rendered object model; and storing each scenario image and the semantic segmentation image corresponding to the scenario image as a set of training data.

Abstract: A device for operation in a wireless network may comprise a transmitter configured to transmit protocol information to a plurality of base stations. The protocol information may comprise resource scheduling information and sequence number information. The resource scheduling information may include a bitmap which indicates whether subchannels are organized consecutively or whether subchannels are distributed. The sequence number information may allow the plurality of base stations to sequentially order received packets. The transmitter may be further configured to transmit data, to the plurality of base stations, in accordance with the resource scheduling information and the sequence number information.

Abstract: Embodiments of the present application provide a lane line processing method and a lane line processing device. The method can include: performing a binarization processing on a first image to obtain a binary image, the first image including lane line points and non-lane line points; performing a connected domain analysis on the binary image to obtain at least one connected domain in the binary image, the connected domain including a plurality of adjacent lane line points; determining lane line points in a group corresponding to a lane line, based on the connected domain; and obtaining representation information of the lane line corresponding to the group, by using the lane line points in the group.

Abstract: In accordance with one aspect of the technology, handwritten works (e.g., notes) produced using computer devices are encoded with digital information that is persistently bound with such works, even after conversion into analog form. In accordance with another aspect of the technology, strokes of artwork or handwriting, or glyphs of computer-rendered text, are encoded in a form that permits any alteration of such work to be detected. A variety of other features and arrangements are also detailed.

Abstract: A computed tomography (CT) image display system (10) includes a mapping unit (32), which receives reconstructed volumetric image data (28) of a subject with values in Hounsfield Units (HU), and a set of reference settings (34), adjusts the set of reference settings (34) to an adjusted set of reference settings (35) according to a pixel-value distribution analysis of the HU values selected according to the reference settings (34), and maps the values in HU to gray scale values according to the adjusted reference settings (35).

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for compressing and decompressing data. In one aspect, a method comprises: processing data using an encoder neural network to generate a latent representation of the data; processing the latent representation of the data using a hyper-encoder neural network to generate a latent representation of an entropy model; generating an entropy encoded representation of the latent representation of the entropy model; generating an entropy encoded representation of the latent representation of the data using the latent representation of the entropy model; and determining a compressed representation of the data from the entropy encoded representations of: (i) the latent representation of the data and (ii) the latent representation of the entropy model used to entropy encode the latent representation of the data.

Abstract: The present disclosure according to at least one embodiment relates to, in the learning process of a child using smart toys, a method and system for providing an interactive service by using a smart toy, which provide more accurate classified emotional state of the child based on at least one or more sensed data items of an optical image, a thermal image, and voice data of the child, as well as adaptively provide a flexible and versatile interactive service according to classified emotions.

Abstract: A face image processing method includes: segmenting a face in an image to be processed to obtain at least one organ image block; respectively inputting the at least one organ image block into at least one first neural network, where at least two different types of organs correspond to different first neural networks; and extracting key point information of an organ from the respective input organ image block by the at least one first neural network to respectively obtain key point information of at least one corresponding organ of the face.

Abstract: A system and method for determining a sentiment, a gender and an age group of a subject in a video while the video is being played back. The video is separated into visual data and audio data, the video data is passed to a video processing pipeline and the audio data is passed to both an acoustic processing pipeline and a textual processing pipeline. The system and method performs, in parallel, a video feature extraction process in the video processing pipeline, an acoustic feature extraction process in the acoustic processing pipeline, and a textual feature extraction process in the textual processing pipeline. The system and method combines a resulting visual feature vector, acoustic feature vector, and a textual feature vector into a single feature vector, and determines the sentiment, the gender and the age group of the subject by applying the single feature vector to a machine learning model.

Abstract: A computer-implemented method using a convolutional neural network is provided. The computer-implemented method includes processing an input image through the convolutional neural network to generate an output image including content features of the input image morphed with style features of a style image. The convolutional neural network includes a feature extraction sub-network, a morpher, and a decoder sub-network.