Abstract: Techniques for active learning-based data labeling are described. An active learning-based data labeling service enables a user to build and manage large, high accuracy datasets for use in various machine learning systems. Machine learning may be used to automate annotation and management of the datasets, increasing efficiency of labeling tasks and reducing the time required to perform labeling. Embodiments utilize active learning techniques to reduce the amount of a dataset that requires manual labeling. As subsets of the dataset are labeled, this label data is used to train a model which can then identify additional objects in the dataset without manual intervention. The process may continue iteratively until the model converges. This enables a dataset to be labeled without requiring each item in the data set to be individually and manually labeled by human labelers.

Abstract: Techniques for model adaptation are described. For example, a method of receiving a call to provide either a model variant or a model variant profile of a deep learning model, the call including desired performance of the deep learning model, a deep learning model identifier, and current edge device characteristics; comparing the received current edge device characteristics to available model variants and profiles based on the desired performance of the deep learning model to generate or select a model variant or profile, the available model variants and profiles determined by the model identifier; and sending the generated or selected model variant or profile to the edge device to use in inference is detailed.

Abstract: Techniques for generating and utilizing machine learning based adaptive instructions for annotation are described. An annotation service can use models to identify edge case data elements predicted to elicit differing annotations from annotators, “bad” data elements predicted to be difficult to annotate, and/or “good” data elements predicted to elicit matching or otherwise high-quality annotations from annotators. These sets of data elements can be automatically incorporated into annotation job instructions provided to annotators, resulting in improved overall annotation results via having efficiently and effectively “trained” the annotators how to perform the annotation task.

Abstract: Techniques for benchmarking a machine learning model/algorithm are described. For example, in some instances a method includes generating an execution plan for benchmarking of at least one task corresponding to a machine learning model based on an identified machine learning model, identified training data, and at least one objective for the benchmarking job; receiving execution statistics about the execution of the task as a part of the benchmarking job according to the execution plan; and updating the execution plan based at least in part on the received execution statistics of the task.

Abstract: Described is a technology by which a computer system operates in a mode that is different from a general purpose operating mode, upon detection of a special actuation mechanism coupled to the computing device. For example, actuation of a special hardware button may boot or resume a sleeping computer system into a direct experience upon actuation, including by launching a special program corresponding to that button. The computer system may thus enter a mode in which it mimics a special purpose device such as a consumer electronics device, e.g., a dedicated media player. When in a direct experience, the computer system may also operate in a constrained/sandboxed mode in which operating system limits available functionality to less than what is available when running as a general purpose computer system, e.g., keyboard operation and/or file access may be limited. Different actuation mechanisms may correspond to different modes.

Abstract: A system and process for harvesting context information from selected content is described. One may use a stylus to indicate what content is to be captured. The context information that may be associated with selected content may include URLs, file names, folder names, text from the content, and ink.

Abstract: Described is a technology by which a computer system operates in a mode that is different from a general purpose operating mode, upon detection of a special actuation mechanism coupled to the computing device. For example, actuation of a special hardware button may boot or resume a sleeping computer system into a direct experience upon actuation, including by launching a special program corresponding to that button. The computer system may thus enter a mode in which it mimics a special purpose device such as a consumer electronics device, e.g., a dedicated media player. When in a direct experience, the computer system may also operate in a constrained/sandboxed mode in which operating system limits available functionality to less than what is available when running as a general purpose computer system, e.g., keyboard operation and/or file access may be limited. Different actuation mechanisms may correspond to different modes.

Abstract: An improved system and method for a context-awareness platform that may be used to adapt an execution environment is provided. A framework with interfaces for setting, tracking, and getting context information about the execution environment of a computer system includes a context engine and a unified data store for aggregating context information at a platform level for access by executables running on the platform. The context engine may receive requests to perform context information operations from executables such as setting, tracking and getting context information in the data store. Upon receiving a notification of a change in context information, one or more executables may change the execution environment such as reconfiguring system settings. In this way, an executable may adapt its execution environment using context information.

Abstract: Described is a technology by which a computer system operates in a mode that is different from a general purpose operating mode, upon detection of a special actuation mechanism coupled to the computing device. For example, actuation of a special hardware button may boot or resume a sleeping computer system into a direct experience upon actuation, including by launching a special program corresponding to that button. The computer system may thus enter a mode in which it mimics a special purpose device such as a consumer electronics device, e.g., a dedicated media player. When in a direct experience, the computer system may also operate in a constrained/sandboxed mode in which operating system limits available functionality to less than what is available when running as a general purpose computer system, e.g., keyboard operation and/or file access may be limited. Different actuation mechanisms may correspond to different modes.

Abstract: In response to a user instruction to initiate media playback, which instruction may come from a dedicated media playback hardware button or from a user interface on a display screen, a computer operating system activates a predesignated media playback user account. The account may be limited to access of media playback applications and files located within shared directories. The operating system also launches a media playback application in response to the user instruction to initiate media playback. Instead of activating an account in response to a media playback instruction, the operating system can execute a media player application within a login screen.

Abstract: Systems, methods, and computer-readable media make media content (e.g., audio, video, or pictorial data) from multiple sources available through a single interface of a client device. The client device may be a portable media playing device (such as a laptop, Tablet PC, MP3 player, portable video player, or the like). By aggregating media content from plural sources (such as a desktop PC, a set top box, etc.) and providing a unified listing of the available media content from these multiple sources on the client device, a user can view all available media content at a single location, without making multiple connections to multiple different sources. Additionally, the user can pull media content from these multiple sources to the client device. Features also are provided to automatically synchronize, obtain, and update media content on the media source(s) and/or the client device. Various ways of handling digital rights management issues associated with copying at least some media content also are described.

Abstract: In a computing device, a method and system for searching for matching ink words or phrases, by comparing a given search term of at least one word (and possibly alternates) with the words in a document, including recognized ink words and any possible alternates for those recognized words as returned by a recognizer. Various matching tests are possible because of the use of alternates, which also may have corresponding probability rankings that may influence the search. Searching may occur in actively edited ink documents, or the recognition results may be saved as saved search file data that can be searched independent of recognition.

Abstract: A method and system for searching digitally annotated computer documents, including searching the annotations. An annotated document may be open for viewing by a user, or alternatively present among closed files that are searched to find files that match a query. In one implementation, the document data and the ink-annotation data exist on two separate layers, with each layer having associated text content, whether as original text or as annotations that are recognized as text. When a user issues a query to search the annotated document, the query is passed to a search engine that performs the pattern matching on the text in each layer, to produce search results. The search engine may produce a search result data structure for each match, which may be sorted with other data structures into a list, such as sorted by pages and coordinates, to present them to a user in a logical ordering.

Abstract: A system and method for associating an ink title with a note is described. A user may hand write a title for a document or note and have the handwritten ink title be displayed when interacting with the system when, for example, viewing the contents of a directory, reviewing the results of a search, viewing a list of most recently used documents, viewing a history of which documents were used, and the like. The use of a handwritten ink title speeds a user's recall of the content of the note over textual titles.

Abstract: A system, method, and user interface for interacting with electronically displayed content is disclosed. In response to a selection of an object, a list of menu options is displayed close to the selected object. In one embodiment, the list of menu options is displayed in a font larger than that used to display the selected object. Through the use of the invention, a user is provided with a technique for actively reading displayed content with minimal distractions from the user interface.

Abstract: A system and method for rendering ink annotations for a displayed image is disclosed. The invention adjusts the thickness of rendered lines for ink annotations when rendered. Ink annotations are created by capturing and connecting a set of data points input by a user. The invention outlines various pixels and weights saturation of anti-aliasing pixels to improve a user's viewing experience.

Abstract: Described is a technology by which a computer system operates in a mode that is different from a general purpose operating mode, upon detection of a special actuation mechanism coupled to the computing device. For example, actuation of a special hardware button may boot or resume a sleeping computer system into a direct experience upon actuation, including by launching a special program corresponding to that button. The computer system may thus enter a mode in which it mimics a special purpose device such as a consumer electronics device, e.g., a dedicated media player. When in a direct experience, the computer system may also operate in a constrained/sandboxed mode in which operating system limits available functionality to less than what is available when running as a general purpose computer system, e.g., keyboard operation and/or file access may be limited. Different actuation mechanisms may correspond to different modes.

Abstract: A system and method for rendering ink annotations for a displayed image is disclosed. The invention adjusts the thickness of rendered lines for ink annotations when rendered. Ink annotations are created by capturing and connecting a set of data points input by a user. The invention outlines various pixels and weights saturation of anti-aliasing pixels to improve a user's viewing experience.

Abstract: In a computing device, a method and system for searching for matching ink words or phrases, by comparing a given search term of at least one word (and possibly alternates) with the words in a document, including recognized ink words and any possible alternates for those recognized words as returned by a recognizer. Various matching tests are possible because of the use of alternates, which also may have corresponding probability rankings that may influence the search. Searching may occur in actively edited ink documents, or the recognition results may be saved as saved search file data that can be searched independent of recognition.

Abstract: In a computing device, a method and system for searching for matching ink words or phrases, by comparing a given search term of at least one word (and possibly alternates) with the words in a document, including recognized ink words and any possible alternates for those recognized words as returned by a recognizer. Various matching tests are possible because of the use of alternates, which also may have corresponding probability rankings that may influence the search. Searching may occur in actively edited ink documents, or the recognition results may be saved as saved search file data that can be searched independent of recognition.