Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using embedded function with a deep network. One of the methods includes receiving an input comprising a plurality of features, wherein each of the features is of a different feature type; processing each of the features using a respective embedding function to generate one or more numeric values, wherein each of the embedding functions operates independently of each other embedding function, and wherein each of the embedding functions is used for features of a respective feature type; processing the numeric values using a deep network to generate a first alternative representation of the input, wherein the deep network is a machine learning model composed of a plurality of levels of non-linear operations; and processing the first alternative representation of the input using a logistic regression classifier to predict a label for the input.

Abstract: Embodiments described herein provide a middle frame image processing schema in a multi-sensor environment, for example improve image processing throughput and snappiness by processing middle frame image data objects of a middle frame image size that is shared amongst a plurality of image sensors. An example apparatus includes a plurality of image sensors, each configured to capture a full frame image data object and output a middle frame image data object. The apparatus further includes an image frame multiplexer communicable with each of the plurality of image sensors, and configured to receive a plurality of middle frame image data objects, and output a selected middle frame image data object from the plurality of middle frame image data objects. The apparatus further includes a processor communicable with the image frame multiplexer, and configured to receive the selected middle frame image data object, and process the selected middle frame image data object.

Abstract: The present disclosure relates to an optical fiber ribbon in which the optical fibers of the optical fiber ribbon are intermittently bonded together at bonding regions along the length of the optical fiber ribbon. The bonding regions of the optical fiber ribbon each include a joining ribbon matrix that have different colors along the length of the optical fiber ribbon.

Abstract: A fixture used in the manufacture of an eyepiece, to cut the eyepiece to a particular shape, and a method of using the fixture to cut the eyepiece to have a desired shape. Embodiments are directed to a configurable fixture to align, hold, and protect a plastic sheet (e.g., a wafer) while a laser cutting apparatus is cutting one or more eyepieces out of the wafer. During the cutting, the fixture protects the eyepieces from reflected laser light by providing voids around the laser cutting lines, and by supporting each eyepiece near its perimeter. The fixture can be quickly rearranged for different eyepieces, different eyepiece shapes, and/or different plastic sheet sizes.

Abstract: Display devices include waveguides with in-coupling optical elements that mitigate re-bounce of in-coupled light to improve overall in-coupling efficiency and/or uniformity. A waveguide receives light from a light source and/or projection optics and includes an in-coupling optical element that in-couples the received light to propagate by total internal reflection in a propagation direction within the waveguide. Once in-coupled into the waveguide the light may undergo re-bounce, in which the light reflects off a waveguide surface and, after the reflection, strikes the in-coupling optical element. Upon striking the in-coupling optical element, the light may be partially absorbed and/or out-coupled by the optical element, thereby effectively reducing the amount of in-coupled light propagating through the waveguide.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

Abstract: The present disclosure provides high bulk tissue products, as well as an apparatus and methods for manufacturing the same. The tissue products provided herein not only have high bulk, but they also have improved surface smoothness, particularly compared to tissue products of similar basis weights.

Abstract: A fiber optic cable assembly comprises first and second cable legs each including a tight buffer surrounding coated optical fibers, and a reduced thickness buffer connecting region, with cable leg being devoid of any surrounding jacket and any tensile strength member. A fiber optic cable assembly devoid of a tensile strength member mechanically coupled to a connector comprises a travel limiting feature that serves to limit travel of the ferrule and inhibit ferrule decoupling when tension is applied to a fiber optic cable.

Abstract: Novel methods and compositions for treating textiles and other materials are disclosed in which persistent odor or other symptoms of biofilm presence can be reduced through the use of compositions comprising N-acetyl cysteine or other agents in combination with certain enzymes.

Abstract: The various embodiments illustrated herein disclose a method for operating an imaging device. the method includes activating a first image sensor at a first duty cycle within a first time period. The method further includes activating a second image sensor at a second duty cycle within the first time period. Additionally, the method includes modifying at least one of the first duty cycle or the second duty cycle based on at least a workflow associated an operator of the imaging device.

Abstract: A method for identifying a model for modeling cable stress relaxation dynamics of an imprint head is provided. The method includes performing a non-contact imprint test run for a predetermined number of wafers. Data from a force trace and a position trace of the imprint head during the non-contact imprint test run are collected to compute cable stress relaxation forces. A model having a plurality of different model orders is generated. A set of parameters of the model for each model order is identified. Residual sum of squares for wafer average error for each of the model orders is calculated based on the obtained cable stress relaxation forces. The set of parameters of one of the plurality of model orders may be based on a difference in the residual sum of squares for wafer average error between the one of the plurality of model orders and a next higher model.

Abstract: Techniques for visualizing text expressions in spreadsheet cells by mapping syntactic elements of the text expressions to two-dimensional (2D) or three-dimensional (3D) configurations of elements, and displaying the 2D or 3D configurations in a graphical user interface, wherein the syntactic relationships between syntactic elements in the text expressions are rendered as spatial relationships between the elements in the 2D or 3D configuration. In one embodiment, a method for converting a text expression into a 2D or 3D configuration includes selecting a spreadsheet cell based on input received from a user input device, wherein the spreadsheet cell comprises a text expression, parsing the text expression, using a logic subsystem, into at least a first syntactic element, mapping the first syntactic element, using the logic subsystem, to a first 2D or 3D element, and displaying the first element in a graphical user interface via a display subsystem.

Abstract: The present disclosure provides high bulk tissue products, as well as an apparatus and methods for manufacturing the same. The tissue products provided herein not only have high bulk, but they also have improved surface smoothness, particularly compared to tissue products of similar basis weights.

Abstract: The current disclosure provides techniques for visualizing text expressions in spreadsheet cells in a more intuitive and user friendly manner by mapping syntactic elements of the text expressions to two-dimensional (2D) configurations of 2D elements, and displaying the 2D configurations in a graphical user interface, wherein the syntactic relationships between syntactic elements in the text expressions are rendered as spatial relationships between the 2D elements in the 2D configuration. In one embodiment, a method for converting a text expression into a 2D configuration comprises selecting a spreadsheet cell based on input received from a user input device, wherein the spreadsheet cell comprises a text expression, parsing the text expression, using a logic subsystem, into at least a first syntactic element, mapping the first syntactic element, using the logic subsystem, to a first two-dimensional (2D) element, and displaying the first 2D element in a graphical user interface via a display subsystem.

Abstract: The various embodiments illustrated herein disclose a method for operating an imaging device. the method includes activating a first image sensor at a first duty cycle within a first time period. The method further includes activating a second image sensor at a second duty cycle within the first time period. Additionally, the method includes modifying at least one of the first duty cycle or the second duty cycle based on at least a workflow associated an operator of the imaging device.

Abstract: A laminated body is disclosed. The laminate body has a leather layer comprising natural leather in which all or a portion of the reticular layer has been removed, a fiber layer comprising natural and/or synthetic fibers positioned at the rear-side of the leather layer, and an adhesive layer between the leather layer and the fiber layer, wherein, in the leather layer, the proportion of the thickness of the reticular layer is 5% or less and the proportion of the thickness of the papillary layer is 95% or more.

Abstract: Embodiments of a tensile strength limiting system are provided. The tensile strength limiting system is configured to cause breakage of an optical fiber cable at a predetermined tensile loading below a tensile strength of the optical fiber cable. The tensile strength limiting system includes a force limiter configured for attachment to the optical fiber cable strung on an aerial pole and a restriction through which the optical fiber cable is configured to be looped. At the predetermined tensile loading, the force limiter is configured to allow the optical fiber cable to pull through the restriction; and the restriction is configured to force the optical fiber cable to bend below a minimum bend radius of a strength member within the optical fiber cable such that the strength member breaks.

Abstract: Products and methods are disclosed relating to a deodorant stick comprising mandelic acid.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using embedded function with a deep network. One of the methods includes receiving an input comprising a plurality of features, wherein each of the features is of a different feature type; processing each of the features using a respective embedding function to generate one or more numeric values, wherein each of the embedding functions operates independently of each other embedding function, and wherein each of the embedding functions is used for features of a respective feature type; processing the numeric values using a deep network to generate a first alternative representation of the input, wherein the deep network is a machine learning model composed of a plurality of levels of non-linear operations; and processing the first alternative representation of the input using a logistic regression classifier to predict a label for the input.

Abstract: Systems and methods are provided for managing data in a grid. In one example approach, a method comprises importing data from a data source to a range of cells in a grid, where the range of cells is specified by a notation input associated with the grid, and where the notation input includes a key specifying the data source and security information associated with the data source.