Abstract: Calibration of gaze tracking equipment is described, for example, in a desktop computing scenario. In various examples, an explicit calibration phase is carried out, optionally followed by an implicit calibration phase. In examples, the explicit calibration phase comprises requesting and receiving user manual input events associated with specified locations and measuring gaze associated with the manual input events. In examples, the implicit calibration phase is carried out without disturbing other activity of a user in the desktop computing environment, such as operating a graphical user interface. In various examples calibration data is stored in a plurality of buffers and used to control switching between explicit and implicit calibration phases.

Abstract: A collection of data that is extremely large can be difficult to search and/or analyze. Relevance may be dramatically improved by automatically classifying queries and web pages in useful categories, and using these classification scores as relevance features. A thorough approach may require building a large number of classifiers, corresponding to the various types of information, activities, and products. Creation of classifiers and schematizers is provided on large data sets. Exercising the classifiers and schematizers on hundreds of millions of items may expose value that is inherent to the data by adding usable meta-data. Some aspects include active labeling exploration, automatic regularization and cold start, scaling with the number of items and the number of classifiers, active featuring, and segmentation and schematization.

Abstract: A distributed messaging system includes a gateway having an interface to receive client messages and having access to a gateway database and to a messaging host that supports multiple partitions that are executed on processors of a cluster of processors. Each of the partitions supports execution of at least one message broker. The gateway database includes a mapping between each of the message brokers and one of the multiple partitions. An admin agent may periodically scan the gateway database to check if the message brokers reach a capacity threshold. The admin agent may create new message brokers and add the new message brokers to a pool of message brokers to allow new queues to be associated with the new message brokers.

Abstract: Systems and methodologies are provided that employ an allocation lock, which permits only a single transaction to acquire space on a particular page at any one time. The allocation lock of the present invention facilitates operations of concurrent transactions at a subpage level (e.g., a row level), and in conjunction with a heap manager can enforce a set of conditions such that prior to a commit stage of a transaction, a space availability for a particular page can be typically assured (e.g., that transactions operating on various copies of the page do not consume all of storage space on that page), and reorganization of data around the page is mitigated (e.g., that a transaction need not move data around the page for purpose of merging various copies.

Abstract: Techniques for separating application processes into a system space and a replay space are described in a record and replay tool. The technique permits applications to run in the replay space while a record and replay library runs and manages resources in system space ensuring accurate replay of saved data that are used by applications.

Abstract: A test document is parsed into components which may include bounding boxes, segments, and points. Test code makes calls to properties and methods of components in order to verify document layout. Rather than take absolute measurements of component placement, components are evaluated relative to each other. Layout verification logic may be part of a larger software test system.

Abstract: A probability distribution for speech model parameters, such as auto-regression parameters, is used to identify a distribution of denoised values from a noisy signal. Under one embodiment, the probability distributions of the speech model parameters and the denoised values are adjusted to improve a variational inference so that the variational inference better approximates the joint probability of the speech model parameters and the denoised values given a noisy signal. In some embodiments, this improvement is performed during an expectation step in an expectation-maximization algorithm. The statistical model can also be used to identify an average spectrum for the clean signal and this average spectrum may be provided to a speech recognizer instead of the estimate of the clean signal.

Abstract: A system and method are provided for defining a projective mapping of quadrilaterals by specifying the images of four independent points and imposing the additional constraint that straight lines are mapped to straight lines. A barycentric coordinate system is used to represent four points in the source space and their corresponding four points in the destination space. The two barycentric coordinate systems are related by a mapping from the points in the source space to the points in the destination space such that the barycentric coordinates of all four input points map to the barycentric coordinates of their respective output points. A ratio of the associated weights is determined for each of the three basis points in the two spaces. These ratios provide a triplet that defines the mapping (transformation) in three coordinates. These three coordinates are then transformed to the 3×3 industry standard projection matrix for projective mapping.

Abstract: A system and method for incremental change synchronization among multiple copies of data is disclosed. To achieve the synchronization a genera synchronization model is used. The synchronization model establishes a one-way data flow path to allow incremental changes to be transferred from a copy of data that incorporates the incremental changes to a copy of data that does not yet incorporate the incremental changes. To achieve this one-way data flow path, the synchronization model uses an agent, a data collector, and a data synchronizer. The data collector collects incremental changes transferred to it and applies the incremental changes to a copy of data. A data synchronizer transfers incremental changes to a data collector. An agent controls the synchronization process by making a connection between a data collector and a data synchronizer and then directing the data synchronizer to transfer data to the data collector.