Microsoft Technology Licensing, LLC

Youjun Liu

Jian Wang

Ji Li

Amit Srivastava

Jiang Wu

Liyong Chen

Yantao Li

Xianfang Wang

Tongxian Chen

Qiang Wang

Yongguang Zhang

Yingnong Dang

Jiahe Helen Wang

Chuanxiong Guo

Qing Yu

Hongyun Yang

Xiaozhi Yu

Gregory Alexander DePaul

The present disclosure relates to processing operations that execute image classification training for domain-specific traffic, where training operations are entirely compliant with data privacy regulations and policies. Image classification model training, as described herein, is configured to classify meaningful image categories in domain-specific scenarios where there is unknown data traffic and strict data compliance requirements that result in privacy-limited image data sets. Iterative image classification training satisfies data compliance requirements through a combination of online image classification training and offline image classification training. This results in tuned image recognition classifiers that have improved accuracy and efficiency over general image recognition classifiers when working with domain-specific data traffic. One or more image recognition classifiers are independently trained and tuned to detect an image class for image classification. Training of independent image recognition classifiers is also utilized for training and tuning of deeper learning models for image classification.

Image classification modeling while maintaining data privacy compliance

A textual user input is received and a plurality of different text-to-content models are run on the textual user input. A selection system attempts to identify a suggested content item, based upon the outputs of the text-to-content models. The selection system first attempts to generate a completed suggestion based on outputs from a single text-to-content model. It then attempts to mix the outputs of the text-to-content models to obtain a completed content suggestion.

Machine learning model-based content processing framework

The present disclosure relates to processing operations configured for an image recognition pipeline that is used to tailor real-time management of image recognition processing for technical scenarios across a plurality of different applications/services. Image recognition processing is optimized at run-time to ensure that latency requirements are met so that image recognition processing results are returned in a timely manner that aids task execution in an application-specific instances. An image recognition pipeline may manage a plurality of image recognition models that comprise a combination of image analysis service (IAS) models and deep learning models. A scheduler of the image recognition pipeline optimizes image recognition processing by selecting at least: a subset of the image recognition models for image recognition processing and a device configuration for execution of the subset of image recognition models, in order to return image recognition results within a threshold time period that satisfies application-specific execution.

Image classification pipeline

A network address mapping system is described. The network address mapping system can identify a set of Web pages, collects information from the Web pages indicating geographical locations ('geolocations'), and correlate the geolocations with the network addresses from which the identified Web pages are served. The collected information can be weighted based on various factors, such as its relative position in a Web page. The collected information can then be used to identify a geolocation. The network mapping system can deduce geolocations for portions of ranges of network addresses based on the score, and can infer geolocations for other portions based on the deduced geolocations. This mapping can then be stored in a database and provided as a geomapping service. The network address mapping system is able to map network addresses to geographical locations. Thereafter, when a user's client computing device accesses a Web server, the Web server can easily and accurately determine a geographical location by querying the database storing the mapping or a geomapping service.

Mapping network addresses to geographical locations

Methods and apparatuses that synchronize a paper document to an associated digital document by establishing a mapping. An embedded interactive code (EIC) Document is created as a digital file that serves as an intermediate tier between the paper document and the digital document. Both the paper document and the EIC document are generated while printing the paper document. The EIC document records the corresponding EIC array allocations and a unique document identification number. An image capturing pen may generate a stroke on any page of paper document. With the EIC document, the methods and apparatuses inform an application the page and location on the page of the stroke.

Embedded interaction code document

A method and apparatus for capturing information encoded within a surface, such as location information or document metadata, and associating the information with a document is described. The captured information may be obtained by a camera associated with a pointing or writing device having a camera, such as an image-capturing pen.

Stroke localization and binding to electronic document

Computer-readable media having computer-executable instructions provide support for the printing of documents containing embedded interactive code information in conjunction with different applications that include Microsoft Word, Microsoft PowerPoint&#xae;, and Microsoft InfoPath&#x2122;. With a Microsoft Word document, the contents and the associated portion of the EIC pattern are printed page by page. A virtual printer intercepts a spooled printed job and saves the associated enhanced metafile. Subsequently, the EMF and the associated EIC pattern for each page are printed. With a Microsoft PowerPoint&#xae; document, each page is printed with a selected number of slides and with an associated portion of the EIC pattern. A note-taking section and play control buttons may be printed. For a Microsoft InfoPath&#x2122; document, an HTML document is obtained from the InfoPath XML document before printing. Microsoft InfoPath prints the image of one page with the associated EIC pattern.

Embedded interaction code printing with Microsoft Office documents

A system and method for editing ink objects recognized in ink input is provided. An ink parser may recognize an ink object in ink input and then an ink editing user interface may edit the ink object recognized by the ink parser. The ink parser may include a chart detector, shape recognizer, and various ink object recognizers such as a chart recognizer, a list detector and a table detector. The various ink object recognizers may recognize particular ink objects. The ink editing user interface may edit the ink object recognized by the ink parser. The ink editing user interface may include a chart editor, list editor, table editor, mode switcher, and a visualizer. The mode switcher may switch the ink editing system between inking mode and ink editing mode.

Growing community of inventors

Beijing, China

Youjun Liu