Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing a machine learning task on a tuple of respective input sequences to generate an output. In one aspect, one of the systems includes a neural network comprising a plurality of encoder neural networks and a head neural network, each encoder neural network configured to: receive a respective input sequence from the tuple; process the respective input sequence using one or more encoder network layers to generate an encoded representation comprising a sequence of tokens; and process each of some or all of the tokens in the sequence of tokens using a projection layer to generate a lower-dimensional representation, and the head neural network configured to: receive lower-dimensional representations of a respective proper subset of the sequence of tokens generated by the encoder neural network; and process the lower-dimensional representations to generate the output.

Abstract: A method manufacturing a spacer for a pipe-in-pipe system includes mixing aerogel particles with a polymer to form a mixture in which the particles are dispersed in the polymer. The resulting mixture is moulded and the polymer is solidified to form the spacer or a component of the spacer, in which the dispersed, particles are suspended in a matrix of the solidified polymer.

Abstract: Described herein are advanced artificial intelligence agents for modeling physical interactions. An apparatus to provide an active artificial intelligence (AI) agent includes at least one database to store physical interaction data and compute cluster coupled to the at least one database. The compute cluster automatically obtains physical interaction data from a data collection module without manual interaction, stores the physical interaction data in the at least one database, and automatically trains diverse sets of machine learning program units to simulate physical interactions with each individual program unit having a different model based on the applied physical interaction data.

Abstract: Skin smoothing is applied to images using a bilateral filter and aided by a skin map. In one example a method includes receiving an image having pixels at an original resolution. The image is buffered. The image is downscaled from the original resolution to a lower resolution. A bilateral filter is applied to pixels of the downscaled image. The filtered pixels of the downscaled image are blended with pixels of the image having the original resolution, and the blended image is produced.

Abstract: A system for performing a machine learning task on a network input is described. The system includes one or more computers and one or more storage devices storing instructions that, when executed by the one or more computers, cause the one or more computers to implement (i) multiple sorting networks in which each sorting network is configured to sort vector blocks in a sequence of vector blocks to generate a sorted sequence of vector blocks; and (ii) a sorting attention neural network configured to perform the machine learning task on the input sequence by executing multiple sorting attention mechanisms using the sorting networks.

Abstract: An embodiment of a computer-implemented method generates a filtered set of region proposals for a digital image processing network. The method comprises generating a keypoint count map for a digital image, wherein the keypoint count map indicates how many keypoints have been detected near different pixels in the digital image. The method also comprises generating a keypoint integral image, wherein the keypoint integral image comprises a summed area table that is based on the keypoint count map. The method also comprises generating object existence indicators for an initial region set for the digital image based on the keypoint integral image, and extracting a keypoint-filtered region set from the initial region set based on the object existence indicators. The method may also comprise using the keypoint-filtered region set as region proposal input for an object proposal subsystem. Other embodiments are described and claimed.

Abstract: A method and apparatus for processing sparse points. The method includes determining spatial hierarchical point data based on a key point set and a local point set of a sparse point set, determining relationship feature data by encoding a spatial hierarchical relationship between points of the spatial hierarchical point data, generating a global feature and a local feature of the sparse point set through a conversion operation associated with the relationship feature data, and generating a processing result for the sparse point set based on the global feature and the local feature.

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: Described herein are advanced artificial intelligence agents for modeling physical interactions. An apparatus to provide an active artificial intelligence (AI) agent includes at least one database to store physical interaction data and compute cluster coupled to the at least one database. The compute cluster automatically obtains physical interaction data from a data collection module without manual interaction, stores the physical interaction data in the at least one database, and automatically trains diverse sets of machine learning program units to simulate physical interactions with each individual program unit having a different model based on the applied physical interaction data.

Abstract: Fiber-reinforced organic polymer aerogels, articles of manufacture and uses thereof are described. The reinforced aerogels include a fiber-reinforced organic polymer matrix having an at least bimodal pore size distribution with a first mode of pores having an average pore size of less than or equal to 50 nanometers (nm) and a second mode of pores having an average pore size of greater than 50 nm and a thermal conductivity of less than or equal to 30 mW/m·K at a temperature of 20° C.

Abstract: Provided is a method of cleaning a collection of resin beads, wherein the method comprises bringing the collection of resin beads into contact with an aqueous solution, wherein the aqueous solution comprises one or more dissolved amine compounds, wherein the collection of resin beads comprises polymer that comprises attached carboxylic acid groups or sulfonic acid groups or a mixture thereof.

Abstract: An implant (500) comprises an elastic deformation portion (51) and a connection portion (52) connected with the proximal end (511) of the elastic deformation portion (51), wherein the elastic deformation portion (51) is covered by an elastic outer layer (55); the proximal end of the connection portion (52) is surrounded by a tightening ring (58); and the proximal end of the elastic outer layer (55) is covered by the tightening ring (58). Since the proximal end of the elastic outer layer (55) of the implant (500) is surrounded by a tightening ring (58), the elastic outer layer (55) is closely attached on the connecting portion (52) of the implant (500), such that the elastic outer layer (55) can be prevented from being rolled over from the surface of the implant (500) when the implant (500) is implanted, thereby improving safety of lung volume reduction surgery.

Abstract: The present application relates to use of hexokinase-2 and a kit for detecting rare tumor cells in body fluid sample. The present application, based on a principle of abnormal tumor cell energy metabolism, conducts a vitro detection by using a glycolysis marker hexokinase-2 (HK2) as a marker, and assists localization and selection of the rare tumor cells by an addressable microporous array chip or a glass sheet, so as to implement a detection on rare tumor cells having high glycolysis activity in a cancer patient's body fluid sample.

Abstract: Various embodiments are generally directed to techniques for employing a hybrid of sequential and parallel processing to perform random sample and consensus (RANSAC). A device to perform RANSAC includes a derivation component to derive a first set of proposed models in parallel from a first set of minimal sample sets of a data set; and a comparison component to recalculate a required quantity of proposed models to derive an accurate model if a proposed model of the first set of proposed models better fits the data set than any proposed model derived prior to derivation of the first set of proposed models, and to determine whether to derive a second set of proposed models following derivation of the first set of proposed models based on a comparison of the required quantity to a quantity of previously derived proposed models that includes the first set. Other embodiments are described and claimed.

Abstract: Antimicrobial silicone-based dressings, such as wound dressings, are disclosed. An example dressing comprises a transparent and self-adhesive gel sheet cured from a liquid containing silicone, the sheet having dispersed therein (i) particulates of a chlorhexidine compound that is not soluble in the liquid; and (ii) at least one other antimicrobial. Methods of making the silicone-based dressings and methods of use are also disclosed.

Abstract: Techniques for using attention events for audience expansion are provided. In one technique, first interaction data that indicates multiple interactions by the first entity with multiple content items is stored. The interactions includes an interaction that is based on an amount of time that content within one of the content items was presented to the first entity. Based on the first interaction data, similarity data that identifies one or more content delivery campaigns that are similar to a particular content delivery campaign is generated. Second interaction data that indicates interaction(s) by a second entity with content item(s) is stored. Based on the second interaction data and the similarity data, association data that associates the second entity with the particular content delivery campaign is stored. The association data may be used to identify the particular campaign in response to receiving a content request from a computing device of the second entity.

Abstract: A pose estimation method includes obtains an event stream from an event-based vision sensor configured to capture a target object to which light-emitting devices flickering at a predetermined first frequency are attached, obtains a polarity change period of at least one pixel based on the event stream, generates an image frame sequence using at least one target pixel having a polarity change period corresponding to the first frequency, among the at least one pixel, extracts a feature sequence including feature vectors corresponding to the at least one target pixel, from the image frame sequence, and estimates a pose sequence of the target object by applying the feature sequence to a deep neural network (DNN) model.

Abstract: Provided is a process for producing 2,2-bis(4-hydroxyphenyl)propane, comprising condensing phenol with acetone in the presence of an acid catalyst; wherein the acid catalyst comprises a collection of sulfonated polymeric beads, wherein the sulfonated polymeric beads comprise (i) 75 to 99% by weight, based on the weight of the bead, polymerized units of monofunctional vinyl monomer, and (ii) 1 to 25% by weight, based on the weight of the bead, polymerized units of multifunctional vinyl monomer; wherein 90% or more of the beads by volume are uniform beads.

Abstract: Techniques for suggesting targeting criteria for a content delivery campaign are provided. An affinity score representing an affinity between the attribute values of each pair of multiple pairs of attribute values is computed. First input indicating a particular attribute value for a particular attribute type is received through a user interface for creating a content delivery campaign. The user interface includes fields for inputting attribute values for multiple attribute types that includes the particular attribute type. In response to the first input and based on affinity scores associated with the particular attribute value, a set of suggested attribute values is identified. The user interface is updated to include the set of suggested attribute values. Second input indicating a selection of a particular suggested attribute value is received. The particular suggested attribute value is added to the content delivery campaign.

Abstract: An embodiment of a computer-implemented method generates a filtered set of region proposals for a digital image processing network. The method comprises generating a keypoint count map for a digital image, wherein the keypoint count map indicates how many keypoints have been detected near different pixels in the digital image. The method also comprises generating a keypoint integral image, wherein the keypoint integral image comprises a summed area table that is based on the keypoint count map. The method also comprises generating object existence indicators for an initial region set for the digital image based on the keypoint integral image, and extracting a keypoint-filtered region set from the initial region set based on the object existence indicators. The method may also comprise using the keypoint-filtered region set as region proposal input for an object proposal subsystem. Other embodiments are described and claimed.