Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for performing machine learning tasks. One method includes receiving (i) a model input, and (ii) data identifying a first machine learning task to be performed on the model input to generate a first type of model output for the model input; augmenting the model input with an identifier for the first machine learning task to generate an augmented model input; and processing the augmented model input using a machine learning model, wherein the machine learning model has been trained on training data to perform a plurality of machine learning tasks including the first machine learning task, and wherein the machine learning model has been configured through training to process the augmented model input to generate a machine learning model output of the first type for the model input.

Abstract: The present disclosure relates to applying techniques similar to those used in neural network language modeling systems to a content recommendation system. For example, by associating consumed media content to words of a language model, the system may provide content predictions based on an ordering. Thus, the systems and techniques described herein may produce enhanced prediction results for recommending content (e.g. word) in a given sequence of consumed content. In addition, the system may account for additional user actions by representing particular actions as punctuation in the language model.

Abstract: A method for mounting a loudspeaker carrier in a vehicle. The method includes providing a loudspeaker carrier and providing an expandable material. The method further includes arranging the expandable material between the loudspeaker carrier and a mounting site on bodywork of the vehicle and activating the expandable material such that the material expands, forming a foam structure, and causes the loudspeaker to be secured to the bodywork.

Abstract: The present disclosure relates to applying techniques similar to those used in neural network language modeling systems to a content recommendation system. For example, by associating consumed media content to words of a language model, the system may provide content predictions based on an ordering. Thus, the systems and techniques described herein may produce enhanced prediction results for recommending content (e.g. word) in a given sequence of consumed content. In addition, the system may account for additional user actions by representing particular actions as punctuation in the language model.

Abstract: Provided are a device and method for securing the ventilator tubing circuitry to a patient so as to prevent the circuitry from becoming disconnected from the patient or ventilator or becoming entangled, crushed or otherwise made non-functional thereby causing possible harm to the patient.

Abstract: A mixture which comprises from 5 to 50% by weight of a block copolymer A, which comprises one or more copolymer blocks (B/S)A each composed of from 65 to 95% by weight of vinylaromatic monomers and from 35 to 5% by weight of dienes and of a glass transition temperature TgA in the range from 40° to 90° C.; from 95 to 50% by weight of a block copolymer B, which comprises at least one hard block S composed of vinylaromatic monomers and comprises one or more copolymer blocks (B/S)B each composed of from 20 to 60% by weight of vinylaromatic monomers and from 80 to 40% by weight of dienes and of a glass transition temperature TgB in the range from ?70° to 0° C.; from 0 to 45% by weight of polystyrene or of a block copolymer C other than A and B; and from 0 to 6% by weight of a plasticizer; and its use for production of shrink films.

Abstract: The present disclosure relates to applying techniques similar to those used in neural network language modeling systems to a content recommendation system. For example, by associating consumed media content to words of a language model, the system may provide content predictions based on an ordering. Thus, the systems and techniques described herein may produce enhanced prediction results for recommending content (e.g. word) in a given sequence of consumed content. In addition, the system may account for additional user actions by representing particular actions as punctuation in the language model.

Abstract: A device for securing a tracheostomy or endotracheal tube to a patient.

Abstract: Examples disclose a method and system for designing voice applications. The method may be executable to receive a verbal input, parse the verbal input to recognize a keyword, and identify a plurality of applications associated with the recognized keyword. Moreover, the method may be further executable to determine a relevance and/or payment associated with the verbal input and/or keyword, to identify one or more applications that are already installed on a computing device, and to initiate or offer to initiate the identified installed application based on the determined relevance and/or payment. When an installed application is not identified, the method may be executable to identify one or more applications from a plurality of relevant candidate applications and present one or more of the identified relevant candidate applications to a user for possible installation. The payment may be based on whether the identified application is already installed on the computing device.

Abstract: A method for mounting a loudspeaker carrier in a vehicle. The method includes providing a loudspeaker carrier and providing an expandable material. The method further includes arranging the expandable material between the loudspeaker carrier and a mounting site on bodywork of the vehicle and activating the expandable material such that the material expands, forming a foam structure, and causes the loudspeaker to be secured to the bodywork.

Abstract: A copolymer containing, in addition to recurring elements of a sulfonated poly(arylene) containing exclusively recurring structural element(s) of the general formulas —[—Ar1(SO3M)n-X—]— and —[—Ar2(SO3M)n-Y—]—, wherein X and Y, which are identical or different from each other, each represent an electron-acceptor group, Ar1 and Ar2, which are either identical or different from each other, represent an aromatic or heteroaromatic ring system with 5-18 ring atoms; wherein the aromatic or heteroaromatic ring system, in addition to the SO3M and the substituents X and Y, optionally comprises additional substituents which are not electron-donor groups; M represents monovalent cation and n is an integral number between 1 and 4; and wherein X, Y, Ar1, Ar2, M and n can be identical or different in various structural elements, independently of each other, one or several additional elements of at least one additional monomer or macromonomer.

Abstract: A mixture which comprises a) from 5 to 95% by weight of a copolymer A which comprises one or more copolymer blocks (B/S)A each composed of from 65 to 95% by weight of vinylaromatic monomers and from 35 to 5% by weight of dienes and of a glass transition temperature TgA in the range from 40° to 90° C., b) from 95 to 5% by weight of a block copolymer B which comprises at least one hard block S composed of from 95 to 100% by weight of vinylaromatic monomers and of from 0 to 5 by weight of dienes, and comprises one or more copolymer blocks (B/S)A each composed of from 65 to 95% by weight of vinylaromatic monomers and of from 35 to 5% by weight of dienes and of a glass transition temperature TgA in the range from 40° to 90° C., and comprises one or more copolymer blocks (B/S)B each composed of from 20 to 60% by weight of vinylaromatic monomers and of from 80 to 40% by weight of dienes and of a glass transition temperature TgB in the range from ?70° to 0° C.

Abstract: A method and system for is disclosed for speech synthesis using deep neural networks. A neural network may be trained to map input phonetic transcriptions of training-time text strings into sequences of acoustic feature vectors, which yield predefined speech waveforms when processed by a signal generation module. The training-time text strings may correspond to written transcriptions of speech carried in the predefined speech waveforms. Subsequent to training, a run-time text string may be translated to a run-time phonetic transcription, which may include a run-time sequence of phonetic-context descriptors, each of which contains a phonetic speech unit, data indicating phonetic context, and data indicating time duration of the respective phonetic speech unit. The trained neural network may then map the run-time sequence of the phonetic-context descriptors to run-time predicted feature vectors, which may in turn be translated into synthesized speech by the signal generation module.

Abstract: A device for securing a tracheostomy or endotracheal tube to a patient.

Abstract: Methods and systems provided can incentivize consuming, sharing, and interacting with online content by providing redeemable points to users for their interaction with content. Points may be redeemed for real or virtual goods and services. Profiles on social networking websites relating to users or their social circles can be mined using natural language processing to determine topical interests relevant to users to facilitate displaying online content most likely to be of interest to users.

Abstract: Hydrolytically and thermo-oxidatively stable sulfonated polyarylenes include the structural element —X—Ar(SO3M)n-Y—. The aromatic ring carrying the sulfonic acid group is substituted exclusively by electron-acceptor bridge groups X and Y and, if applicable, by other non-electron-donor substituents. Their synthesis and application are also included.

Abstract: A filter, in particular for compressed gas systems, includes a filter top part having at least one inlet channel and at least one outlet channel and a filter bottom part having a filter element, wherein the filter top part has a pressure-resistant housing and an insert which is suited in an optimal manner to the course of the inlet and outlet channels.

Abstract: A speaker assembly arrangement for a vehicle is provided that includes a speaker unit including a speaker cabinet defining a first volume and a body part of a vehicle including a hollow member defining a second volume. The speaker cabinet is acoustically coupled to the hollow member by a press-fit coupling having a press-fit direction and mounted on the body part by at least one fastener having a fastening direction substantially perpendicular to the press-fit direction. The body part comprises a first guide arranged and configured to slidably engage with a second guide arranged on the speaker cabinet so that force on the speaker cabinet in the fastening direction is at least in part redirected to the press-fit direction by the slidable engagement of the first guide and the second guide.

Abstract: The invention relates to a waste water immersion probe (10) comprising a liquid-tight housing (11), a sensor (18) which is disposed in the housing (11) and a sensor window (20) formed on the housing (11). A wiper element (22) for cleaning the sensor window (20) is arranged on the outside of the sensor window (20), and the wiper element (22) is driven by a drive motor (24) which is arranged essentially in the housing (11). The housing is devoid of openings. A magnetic element (36, 37) is associated with the drive motor (24) and a magnetic element (34, 35) is associated with the wiper element (22) in such a manner that the magnetic drive motor element (36, 37) produces a magnetic field which transverses the housing wall and transmits a drive torque to the wiper element (22) without a shaft.