Abstract: The performance of a neural network (NN) and/or deep neural network (DNN) can limited by the number of operations being performed as well as management of data among the various memory components of the NN/DNN. A sparsity-inducing regularization optimization process is performed on a machine learning model to generate a compressed machine learning model. A machine learning model is trained using a first set of training data. A sparsity-inducing regularization optimization process is executed on the machine learning model. Based on the sparsity-inducing regularization optimization process, a compressed machine learning model is received. The compressed machine learning model is executed to generate one or more outputs.

Abstract: The performance of a neural network (NN) and/or deep neural network (DNN) can limited by the number of operations being performed as well as management of data among the various memory components of the NN/DNN. A sparsity-inducing regularization optimization process is performed on a machine learning model to generate a compressed machine learning model. A machine learning model is trained using a first set of training data. A sparsity-inducing regularization optimization process is executed on the machine learning model. Based on the sparsity-inducing regularization optimization process, a compressed machine learning model is received. The compressed machine learning model is executed to generate one or more outputs.

Abstract: The performance of a neural network (NN) and/or deep neural network (DNN) can limited by the number of operations being performed as well as management of data among the various memory components of the NN/DNN. A sparsity-inducing regularization optimization process is performed on a machine learning model to generate a compressed machine learning model. A machine learning model is trained using a first set of training data. A sparsity-inducing regularization optimization process is executed on the machine learning model. Based on the sparsity-inducing regularization optimization process, a compressed machine learning model is received. The compressed machine learning model is executed to generate one or more outputs.

Abstract: A facility enabling a user to operate visual user interface controls with ink commands is described. The facility causes to be displayed a control operable with mouse input, and receives an ink command directed to the displayed control. In response, the facility operates the displayed control in accordance with the received ink command.

Abstract: Described herein is a system and method for visually enhancing digital ink of an electronic document. A trigger to visually enhance digital ink of portion(s) of the electronic document is received. In response to the received trigger, the digital ink of portion(s) of the electronic document to determine a semantic structure of the digital ink in response to the received trigger. The digital ink of the portion(s) of the electronic document are visually enhanced in accordance with the determined semantic structure. Visual enhancement can include horizontal line adjustment, aligning line(s), aligning word in a particular line using a baseline, adjusting vertical spacing of lines, paragraphs, and/or lists, adjusting spacing between words and/or list items in a particular line, modifying ink styling (e.g., ink size, ink thickness, ink color), adjusting sizing of characters in a same group, unifying ink color, and/or unifying ink thickness.

Abstract: A computing device includes a digital input system that allows freehand digital inputs to be received (e.g., via movement of a pen, stylus, finger, etc.). The digital input system provides functionality allowing applications to receive digital inputs from a user and control the display of data based on the digital inputs. The digital input system receives digital input from a user and analyzes the digital input to collect input data for the digital input. As the digital input is received, the input data is made available to an application, allowing the application to operate on the input data (e.g., modify the input data and/or generate additional input data) and provide the operated-on input data to the digital input system. The digital input system then proceeds to control the display of data based on the operated-on input data.

Abstract: A facility for adapting the prediction of ink is described. In some examples, the facility receives information about a spatial movement by a user. On the basis of the received information, the facility predicts future spatial movement by the user, and generates an ink stroke that reflects both the spatial movement described by the received information and at least a portion of the predicted future spatial movement. The facility enforces against the generated ink stroke a limit that has the effect of controlling the area of a portion of the ink stroke corresponding to the at least a portion of the predicted future spatial movement, and causes the generated ink stroke, subject to the enforcement of the limit, to be displayed.

Abstract: Digital ink is generated to represent a visual component, such as a letter, number, character, and/or other symbol. The digital ink is generated by obtaining multiple different curves that combine to generate the visual component. These different curves can have various different characteristics (e.g., different thicknesses) to provide the desired visual component. The combined curves are converted into a set of primitives that make up the parts of the combined curves, and the set of primitives are converted into a digital ink format. Data describing the set of primitives in digital ink format can be stored and subsequently used to display the visual component as digital ink.

Abstract: An application on a computing device includes a digital ink system that supports the input and output of digital ink in a markup language document. Digital ink refers to a digital representation of object (e.g., pen or finger) strokes on an input device that can be displayed on an output device. The markup language document can include, for example, an edit box in which the digital ink and optionally other data can be input. As the input object moves over the input device, data identifying the locations of the input object are captured and embedded in a markup language element of the document. The digital ink system also allows documents including digital ink to be displayed, the digital ink system displaying the digital ink along with any other data included in the document.

Abstract: Platform-specific user interface (UI) objects may be generated based on received free hand input. In response to receiving free hand input, the received free hand input is analyzed. Based on the analysis of the received free hand input, one or more elements associated with the free hand input are identified. The one or more elements can comprise shapes or text corresponding to the received free hand input. For each of at least one of the one or more elements, the one or more elements are analyzed. Analyzing the at least one element may comprise analyzing a context of the at least one element. A UI object corresponding to the at least one element is determined. Executable platform-specific UI code associated with the determined UI object is generated.

Abstract: Methods, systems, and apparatuses for beautifying raw drawing data include a beautification engine having a raw data processor that receives the drawing data and identifies a set of segments in the data. A beautification director constructs a structural representation of the set of segments data that specifies a topological relationship between the segments. A single primitive beautifier identifies a segment in the structural representation to be beautified and a single primitive beautification action is applied to the segment to generate beautified raw drawing data. A multiple primitive beautifier identifies adjacent segments in the structural representation to be beautified and a multiple beautification action is applied to the adjacent segments to generate further beautified drawing data.

Abstract: Systems and methods are provided for improving the latency for display of ink during user creation of ink content with an object, such as a stylus, mouse, finger (or other touch input), or other drawing device. In order to reduce or minimize the time for display of ink content created by a user, aspects of the technology described herein generate predictive wet ink that can be displayed with actual wet ink. The predictive wet ink is calculated by extending an active ink segment a predictive distance. The predictive distance is intended to be a distance between a termination ink point for an active wet ink segment and a tip of the writing object. In one aspect, the predictive distance is calculated by determining the current latency period on a touch device and the velocity of the writing object.

Abstract: A facility enabling a user to operate visual user interface controls with ink commands is described. The facility causes to be displayed a control operable with mouse input, and receives an ink command directed to the displayed control. In response, the facility operates the displayed control in accordance with the received ink command.

Abstract: A facility for adapting the prediction of ink is described. In some examples, the facility receives information about a spatial movement by a user. On the basis of the received information, the facility predicts future spatial movement by the user, and generates an ink stroke that reflects both the spatial movement described by the received information and at least a portion of the predicted future spatial movement. The facility enforces against the generated ink stroke a limit that has the effect of controlling the area of a portion of the ink stroke corresponding to the at least a portion of the predicted future spatial movement, and causes the generated ink stroke, subject to the enforcement of the limit, to be displayed.

Abstract: Symbol recognition techniques may be applied to documents comprising various forms of content. Documents including both text and mathematical expressions may be problematic, as applying a recognizer that does not match the content may produce anomalous results. Instead, a parser may evaluate the document to classify respective regions as one of a text region or a mathematics region, based on the characteristics of each type of content. The recognizer corresponding to the content of each region may be applied to produce a composite document comprising both recognized text expressions and recognized mathematical expressions. Additional functionality may be presented based on the recognized content; e.g.

Abstract: The inertia system provides a common platform and application-programming interface (API) for applications to extend the input received from various multi-touch hardware devices to simulate real-world behavior of application objects. To move naturally, application objects should exhibit physical characteristics such as elasticity and deceleration. When a user lifts all contacts from an object, the inertia system provides additional manipulation events to the application so that the application can handle the events as if the user was still moving the object with touch. The inertia system generates the events based on a simulation of the behavior of the objects. If the user moves an object into another object, the inertia system simulates the boundary characteristics of the objects. Thus, the inertia system provides more realistic movement for application objects manipulated using multi-touch hardware and the API provides a consistent feel to manipulations across applications.

Abstract: A computing device includes a digital input system that provides digital input functionality for the computing device. The digital input system receives both a first input and a second input. The first input is typically an input with the user's dominant hand (e.g., using a pen, stylus, finger, etc.), and the second input is typically an input with the user's non-dominant hand. The digital input system determines whether a second input is received simultaneously with the first input, and selects one of multiple behaviors for the first input based on whether the second input is received simultaneously with the first input. Various different behaviors for the first input can be selected from, such as freehand or freeform input, constraining digital input corresponding to the first input to be along an edge of a stencil, and so forth.

Abstract: Methods for automatic language detection for handwritten text are performed by systems and devices. Such automatic language detection is performed prior to sending representations of the handwritten text to a language recognition engine. Handwritten inputs including one or more writing strokes are received from an input interface, and are associated with coordinates of the inputs and times that the inputs are made. The handwritten inputs are grouped into words based on the coordinates and times. Writing strokes are normalized, and then the words are individually transformed to generate language vectors, such as through a recurrent neural network. The language vectors are used to determine language probabilities for the handwritten inputs. Based on the language probabilities, the handwritten inputs are provided to a specific language recognition engine to determine the language thereof prior to translation or transcription.

Abstract: Described herein is a system and method for visually enhancing digital ink of an electronic document. A trigger to visually enhance digital ink of portion(s) of the electronic document is received. In response to the received trigger, the digital ink of portion(s) of the electronic document to determine a semantic structure of the digital ink in response to the received trigger. The digital ink of the portion(s) of the electronic document are visually enhanced in accordance with the determined semantic structure. Visual enhancement can include horizontal line adjustment, aligning line(s), aligning word in a particular line using a baseline, adjusting vertical spacing of lines, paragraphs, and/or lists, adjusting spacing between words and/or list items in a particular line, modifying ink styling (e.g., ink size, ink thickness, ink color), adjusting sizing of characters in a same group, unifying ink color, and/or unifying ink thickness.

Abstract: A facility for adapting the prediction of ink is described. In some examples, the facility receives information about a spatial movement by a user. On the basis of the received information, the facility predicts future spatial movement by the user, and generates an ink stroke that reflects both the spatial movement described by the received information and at least a portion of the predicted future spatial movement. The facility enforces against the generated ink stroke a limit that has the effect of controlling the area of a portion of the ink stroke corresponding to the at least a portion of the predicted future spatial movement, and causes the generated ink stroke, subject to the enforcement of the limit, to be displayed.