Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a target sequence from an input sequence. In one aspect, a method comprises maintaining a set of current hypotheses, wherein each current hypothesis comprises an input prefix and an output prefix. For each possible combination of input and output prefix length, the method extends any current hypothesis that could reach the possible combination to generate respective extended hypotheses for each such current hypothesis; determines a respective direct score for each extended hypothesis using a direct model; determines a first number of highest-scoring hypotheses according to the direct scores; rescores the first number of highest-scoring hypotheses using a noisy channel model to generate a reduced number of hypotheses; and adds the reduced number of hypotheses to the set of current hypotheses.

Abstract: A fluid end assembly of a hydraulic fluid pump includes a housing having a bore, a removable valve cover closing the bore, and a retainer engaging the bore. The retainer includes a set screw bore that extends from a first side of the retainer to a second side of the retainer, opposite the first side. The fluid end assembly further includes a set screw. The set screw is threaded into the set screw bore from the first side of the retainer. The set screw is not removable from the second side of the retainer. The first side of the retainer is positioned adjacent to the valve cover.

Abstract: A fluid end assembly of a hydraulic fluid pump includes a housing having an axial bore, a removable valve cover closing the axial bore, and a retainer engaging the axial bore. The retainer includes a central opening. The fluid end assembly further includes a threaded drive member positioned within the central opening and operable to apply a preload against the retainer.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a target sequence from an input sequence. In one aspect, a method comprises maintaining a set of current hypotheses, wherein each current hypothesis comprises an input prefix and an output prefix. For each possible combination of input and output prefix length, the method extends any current hypothesis that could reach the possible combination to generate respective extended hypotheses for each such current hypothesis; determines a respective direct score for each extended hypothesis using a direct model; determines a first number of highest-scoring hypotheses according to the direct scores; rescores the first number of highest-scoring hypotheses using a noisy channel model to generate a reduced number of hypotheses; and adds the reduced number of hypotheses to the set of current hypotheses.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a target sequence from an input sequence. In one aspect, a method comprises maintaining a set of current hypotheses, wherein each current hypothesis comprises an input prefix and an output prefix. For each possible combination of input and output prefix length, the method extends any current hypothesis that could reach the possible combination to generate respective extended hypotheses for each such current hypothesis; determines a respective direct score for each extended hypothesis using a direct model; determines a first number of highest-scoring hypotheses according to the direct scores; rescores the first number of highest-scoring hypotheses using a noisy channel model to generate a reduced number of hypotheses; and adds the reduced number of hypotheses to the set of current hypotheses.

Abstract: Technologies are disclosed herein for statistical machine translation. In particular, the disclosed technologies include extensions to conventional machine translation pipelines: the use of multiple domain-specific and non-domain-specific dynamic language translation models and language models; cluster-based language models; and large-scale discriminative training. Incremental update technologies are also disclosed for use in updating a machine translation system in four areas: word alignment; translation modeling; language modeling; and parameter estimation. A mechanism is also disclosed for training and utilizing a runtime machine translation quality classifier for estimating the quality of machine translations without the benefit of reference translations.

Abstract: A well servicing pump includes a power end and a fluid end operably coupled to the power end. The fluid end includes a casing that defines at least one pumping chamber, the pumping chamber at least partially defined by a spherical wall, and a first cylindrical wall extending along a first axis and defining a portion of a power end bore, the first cylindrical wall intersecting the spherical wall to define a first interface edge. A second cylindrical wall extends along a second axis and defines a portion of a discharge bore, the second cylindrical wall intersecting the spherical wall to define a second interface edge, and a third cylindrical wall extends along a third axis and defines a portion of a suction bore, the third cylindrical wall intersecting the spherical wall to define a third interface edge, wherein the first axis, the second axis, and the third axis are arranged between 100 and 140 degrees with respect to one another.

Abstract: A method of processing geophysical data from a survey of a surveyed region of the earth to provide a representation of the underlying geology of said surveyed region, the method comprising inputting 2-D seismic data for said surveyed region, generating a first representation depicting a plurality of reflection interfaces of said underlying geology of said surveyed region using said input seismic data, inputting potential field data for said surveyed region, generating a second representation depicting a plurality of density interfaces of said underlying geology of said surveyed region using said input potential field data; combining said representation of said plurality of density interfaces with said representation of said plurality of reflection interfaces to reposition at least one of said plurality of reflection interfaces and generating a final representation depicting said underlying geology of said surveyed region using said combined first and second representations.

Abstract: Methods, systems, and apparatus, including computer program products, for language translation are disclosed. In one aspect, a method includes accessing a translation hypergraph that represents a plurality of candidate translations, the translation hypergraph including a plurality of paths including nodes connected by edges; calculating first posterior probabilities for each edge in the translation hypergraph; calculating second posterior probabilities for each n-gram represented in the translation hypergraph based on the first posterior probabilities; and performing decoding on the translation hypergraph using the second posterior probabilities to convert a sample text from a first language to a second language.

Abstract: Methods, systems, and apparatus, including computer program products, for language translation are disclosed. In one aspect, a method includes accessing a translation hypergraph that represents a plurality of candidate translations, the translation hypergraph including a plurality of paths including nodes connected by edges; calculating first posterior probabilities for each edge in the translation hypergraph; calculating second posterior probabilities for each n-gram represented in the translation hypergraph based on the first posterior probabilities; and performing decoding on the translation hypergraph using the second posterior probabilities to convert a sample text from a first language to a second language.

Abstract: A ball-coupling and protection apparatus and system is disclosed. In a typical embodiment, a coupler apparatus is disposed along an outer perimeter of a racket, typically a location along the perimeter farthest from the user's grip on the racket. In a typical embodiment, the coupler is a set of hooks from a hook and loop fastener set.

Abstract: An optical fiber management assembly comprises a support module (12) carrying two or more pivotable optical fiber storage trays (13). The support module has channels in its outer surface for guiding fibers to an entry zone in a first region of its surface to a tray selection zone from which fibers are guided to respective trays.

Abstract: A method and system of evaluating crop performance facilitates characterization of the environmental impact of a geographic region or areas within the region for growing plant-life. Environmental measurements are obtained. The environmental measurements are associated with a geographic region. Each environmental measurement includes at least one of soil data and weather data. Respective location data is obtained. The location data is associated with corresponding environmental measurements. An estimated performance characteristic is determined for a particular crop planted in the geographic region based on the obtained environmental measurements and respective location data. Contours are established for one or more uniform performance areas with generally uniform performance characteristic within the geographic region by applying a decision-tree analysis to the obtained environmental measurements.

Abstract: A real-time performance monitoring system for monitoring an electric power grid. The electric power grid has a plurality of grid portions, each grid portion corresponding to one of a plurality of control areas. The real-time performance monitoring system includes a monitor computer for monitoring at least one of reliability metrics, generation metrics, transmission metrics, suppliers metrics, grid infrastructure security metrics, and markets metrics for the electric power grid. The data for metrics being monitored by the monitor computer are stored in a data base, and a visualization of the metrics is displayed on at least one display computer having a monitor. The at least one display computer in one said control area enables an operator to monitor the grid portion corresponding to a different said control area.

Abstract: An optical fibre management assembly comprises a support module (12) carrying two or more pivotable optical fibre storage trays (13). The support module has channels in its outer surface for guiding fibres to an entry zone in a first region of its surface to a tray selection zone from which fibres are guided to respective trays.

Abstract: The present invention comprises a method of making metallic fibers and electrodes. The metallic fibers are manufactured by milling a stock piece of material, preferably with a computer number controlled (CNC) milling machine. According to one embodiment, the length of the fibers is determined by the width of the stock material and stock material may be stacked so as to produce a number of fibers. According to another embodiment, the cutting tool comprises chip breakers which determine the length of the fibers. Zinc fibers made according to the present invention may be used to manufacture battery electrodes by compressing the fibers into the desired shape and porosity.

Abstract: A manager loadable printer, an agent that communicates with a printer, and related methods for enhancing printer functionality.