Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: A heat exchanger includes a fluid flow tube configured to allow an exchange of heat between the fluid flowing in the tube and an enclosure surrounding the tube. The enclosure is formed from two profiled bars configured to surround the tube over at least one part of its length. Each of the bars includes a recess adapted to receive the tube, and a joint surface configured to form a joint with the other profiled bar. The joint surfaces of the profiled bars form a gap therebetween. As well, the joint surfaces of the profiled bars are arranged in different planes such that the gap forms at least one baffle on each side of the tube.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for content recommendation using neural networks. In One aspect, a method includes: receiving context information for an action recommendation from multiple possible actions; processing the context information using a neural network that includes Bayesian neural network layers to generate, for each of the actions, one or more parameters of a distribution over possible action scores for the action, where each parameter for each Bayesian layer is associated with data representing a probability distribution over multiple possible current values for the parameter; for each parameter of each Bayesian neural network layer, selecting the current value for the parameter using data representing probability distribution over possible current values for the parameter; and selecting an action from multiple possible actions using the parameters of the distributions over the possible action scores for the action.

Abstract: Heat exchanger assembly comprising a ram air flow channel (14) extending in a longitudinal direction, and characterized in that said assembly comprises: at least two separate heat exchangers (12a, 12b) that are adjacent in a transverse direction perpendicular to the longitudinal direction, are arranged in the ram air flow channel (14), and are configured such that the ram air passing through said channel (14) forms a cold pass of each heat exchanger (12a, 12b) by passing through said heat exchanger (12a, 12b) in said longitudinal direction, each heat exchanger (12a, 12b) also being configured for the passage therethrough of a fluid that is intended to be cooled and that forms a hot pass (20a, 20b); and an air passage which is provided between the heat exchangers and forms a thermally insulating air gap (18) between said exchangers (12a, 12b), and through which the ram air flows, said air passage extending in said longitudinal direction of said ram air flow channel (14).

Abstract: User interfaces for browsing and/or searching are described. In at least certain embodiments, icon data from a plug-in includes badge data and an image (such as a thumbnail). The badge data includes additional information on the file, the creator application of the file, and/or a third party vendor of the creator application. For example, the badge data may include the version of the creator application. Using the badge data, a badge may be generated and displayed on the icon. Other embodiments are also described, and computer readable media and apparatuses are also described.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for content recommendation using neural networks. One of the methods includes receiving context information for an action recommendation; processing the context information using a neural network that comprises one or more Bayesian neural network layers to generate, for each of the actions, one or more parameters of a distribution over possible action scores for the action and selecting an action from plurality of possible actions using the parameters of the distributions over the possible action scores for the action.

Abstract: User interfaces for browsing and/or searching are described. In one embodiment, a method includes displaying a first display area for display representations of documents matching a search query, the first display area configured to display content of the documents which can have a plurality of different types of content including at least one of text-based content and a folder, and displaying a second display area for selecting a selected document to be displayed in the first display area. Other embodiments are also described, and computer readable media and apparatuses are also described.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and machine readable tangible storage media that can provide for smooth and seamless opening of a file from, for example, a preview are described. A preview of a file can be generated by a non-native viewer or reader application and in response to a command to open a file in a native application, the preview can continue to be displayed while the native application launches such that a user will perceive no interruption in display of the content. The non-native application can pass a display state (e.g. a page number or scroll position) to the native application such that the native application continues to show a user's current position or view port into the file.

Abstract: Embodiments of tokenized search suggestions include generating, by at least one suggestion engine executing on a data processing system, first suggestions based on first text input in a text input field. The first suggestions are displayed on a display device and a selection of one of the first suggestions is received. The first text input is replaced with a first token corresponding to the selected one of the first suggestions. The at least one suggestion engine generates second suggestions based on second text input in the text input field and filters the second suggestions using the selected one of the first suggestions. The filtered selections are displayed on a display device.

Abstract: Apparatuses and methods relating to content sharing are described. In one embodiment, a match between content type and one or more content services determine what content sharing services are displayed. In one embodiment, content sharing is accessible to all applications in an operating system.

Abstract: We describe a signal processor, the signal processor comprising: a probability vector generation system, wherein said probability vector generation system has an input to receive a category vector for a category of output example and an output to provide a probability vector for said category of output example, wherein said output example comprises a set of data points, and wherein said probability vector defines a probability of each of said set of data points for said category of output example; a memory storing a plurality of said category vectors, one for each of a plurality of said categories of output example; and a stochastic selector to select a said stored category of output example for presentation of the corresponding category vector to said probability vector generation system; wherein said signal processor is configured to output data for an output example corresponding to said selected stored category.

Abstract: Embodiments of tokenized search suggestions include generating, by at least one suggestion engine executing on a data processing system, first suggestions based on first text input in a text input flea The first suggestions are displayed on a display device and a selection of one of the first suggestions is received. The first text input is replaced with a first token corresponding to the selected one of the first suggestions. The at least one suggestion engine generates second suggestions based on second text input in the text input field and filters the second suggestions using the selected one of the first suggestions. The filtered selections are displayed on a display device.

Abstract: User interfaces for browsing and/or searching are described. In at least certain embodiments, icon data from a plug-in includes badge data and an image (such as a thumbnail). The badge data includes additional information on the file, the creator application of the file, and/or a third party vendor of the creator application. For example, the badge data may include the version of the creator application. Using the badge data, a badge may be generated and displayed on the icon. Other embodiments are also described, and computer readable media and apparatuses are also described.

Abstract: Embodiments of tokenized search suggestions include generating, by at least one suggestion engine executing on a data processing system, first suggestions based on first text input in a text input field. The first suggestions are displayed on a display device and a selection of one of the first suggestions is received. The first text input is replaced with a first token corresponding to the selected one of the first suggestions. The at least one suggestion engine generates second suggestions based on second text input in the text input field and filters the second suggestions using the selected one of the first suggestions. The filtered selections are displayed on a display device.

Abstract: Methods, machine media, and systems that can manipulate preview panels are disclosed. In one embodiment, a preview panel can be transformed, in response to user interaction with the preview panel, into an icon representing the file containing the content displayed in the preview panel. The preview panel can, in one embodiment, be opened from a user interaction with a first icon and the preview panel can be transformed into a second icon that represents the file and can be manipulated as if it were the first icon.