118results about How to "High dimensional" patented technology

A plurality of standard cells 10 are arranged to form a channel-less standard cell array 1, which has vertical and horizontal sides. A plurality of first proximity dummy cells 20 are arranged along each of the vertical sides of the standard cell array to form a first proximity dummy bands 20 such that the upper and lower sides of the first proximity dummy cells are in contact with each other and such that the left or right side of each of the first proximity dummy cells is in contact with the vertical side of the standard cell array 1. Furthermore, a plurality of second proximity dummy bands are arranged along each of the horizontal sides of the standard cell array to form a second proximity dummy bands such that the upper or lower side of each of the second proximity dummy cells is in contact with the horizontal side of the standard cell 1.

A procedure for clustering documents that operates in high dimensions, processes tens of thousands of documents and groups them into several thousand clusters or, by varying a single parameter, into a few dozen clusters. The procedure is specified in two parts: computing a similarity score representing the k most similar documents (typically the top ten) for each document in the collection, and grouping the documents into clusters using the similarly scores.

A plurality of standard cells 10 are arranged to form a channel-less standard cell array 1, which has vertical and horizontal sides. A plurality of first proximity dummy cells 20 are arranged along each of the vertical sides of the standard cell array to form a first proximity dummy bands 20 such that the upper and lower sides of the first proximity dummy cells are in contact with each other and such that the left or right side of each of the first proximity dummy cells is in contact with the vertical side of the standard cell array 1. Furthermore, a plurality of second proximity dummy bands are arranged along each of the horizontal sides of the standard cell array to form a second proximity dummy bands such that the upper or lower side of each of the second proximity dummy cells is in contact with the horizontal side of the standard cell 1.

Methods and systems are disclosed for autonomously building a predictive model of outcomes. A most-predictive set of signals Sk is identified out of a set of signals s1, s2, . . . , sD for each of one or more outcomes ok. A set of probabilistic predictive models ôk=Mk (Sk) is autonomously learned, where ôk is a prediction of outcome ok derived from the model Mk that uses as inputs values obtained from the set of signals Sk. The step of autonomously learning is repeated incrementally from data that contains examples of values of signals s1, s2, . . . , sD and corresponding outcomes o1, o2, . . . , oK. Various embodiments are also disclosed that apply predictive models to various physiological events and to autonomous robotic navigation.

A method for data-driven anomaly detection may include monitoring a plurality of parameters associated with a plurality of subsystems of a system. The method may also include collecting data corresponding to each of the plurality of parameters from the plurality of subsystems and generating monitoring quantities based on the data. The method may also include determining if any quantities in the monitoring quantities exceed a predetermined limit. A contribution plot may be generated corresponding to each of the parameters in response to any of the quantities exceeding the predetermined limit. The method may further include determining which parameter is likely to cause an effect based on the contribution plot.

Embodiments of the invention relate to improvements to the support vector machine (SVM) classification model. When text data is significantly unbalanced (i.e., positive and negative labeled data are in disproportion), the classification quality of standard SVM deteriorates. Embodiments of the invention are directed to a weighted proximal SVM (WPSVM) model that achieves substantially the same accuracy as the traditional SVM model while requiring significantly less computational time. A weighted proximal SVM (WPSVM) model in accordance with embodiments of the invention may include a weight for each training error and a method for estimating the weights, which automatically solves the unbalanced data problem. And, instead of solving the optimization problem via the KKT (Karush-Kuhn-Tucker) conditions and the Sherman-Morrison-Woodbury formula, embodiments of the invention use an iterative algorithm to solve an unconstrained optimization problem, which makes WPSVM suitable for classifying relatively high dimensional data.

The present invention provides an electrode plate having a collector 1, an active material layer 2 and a non-coated portion 3 and satisfies at least one of the following conditions. (1) The maximum thickness in an area inside by 20 mm from an edge portion of the active material layer is not more than the sum of an average thickness of the active material layer and 10 mum. (2) In an inclined portion of the peripheral edge portion of the active material layer, an area in which the thickness of the active material layer is not less than 1 mum and less than an average thickness of the active material layer has a width of not more than 1 mm. (3) An absolute value of a positional shifting of an actually formed pattern with respect to a true boundary line of a predetermined pattern is not more than 1 mm. (4) The active material layers are formed in a plane symmetry to front and back surfaces of the collector, and an absolute value in the positional shifting between the patterns formed on front and back side surfaces of the collector is not more than 1 mm.

A positive resist composition comprises: (A) a resin that contains a repeating unit (A1) having a lactone structure and a cyano group, and increases its solubility to an alkali developer by action of an acid; (B) a compound that generates an acid by irradiation with actinic ray or radiation; and (C) a solvent.

A printer comprising a stationary printhead extending across a paper path is provided. The printhead has a plurality of color channels, each color channel comprising at least one nozzle row extending along the length of the printhead. Each nozzle in a color channel ejects the same colored ink. The printhead has a print zone of less than 10 mm, where the print zone is deifined as the distance covered by all nozzle rows of the printhead in a paper feed direction.

A plate formed with a plurality of regular projections projecting from both faces of the plate can suitably be used as a separator for a fuel cell. Projections on one face of the plate and projections on the opposite face of the plate are alternately and adjacently formed in at least one direction along a plane of a reference plate portion of a metal plate. The reference plate portion has not been deformed. The top face sections of the projections or bottom face sections of the projections are thicker than their peripheral corner portions.

The present invention provides a magnetic memory device capable of performing stable operation efficiently using a magnetic field generated by write current and formed with high precision while realizing a compact configuration. Since a plating film is used for at least a part of a magnetic yoke, as compared with the case of formation by a dry film forming method, sufficient thickness and higher dimensional precision can be obtained. Consequently, a more stabilized return magnetic field can be generated and high reliability can be assured. Neighboring memory cells can be disposed at narrower intervals, so that the invention is suitable for realizing higher integration and higher packing density.

A photosensitive composition for forming a dielectric of the present invention comprising inorganic particles, an alkali developable resin and additives, wherein the additives comprise a compound having a quinonediazido group (C1), a compound containing at least two alkyletherified amino groups in the molecule (C2) and a thermal acid generator (C3), or wherein the inorganic particles comprise inorganic superfine particles (A-I) having a mean particle diameter of less than 0.05 mum and inorganic fine particles (A-II) having a mean particle diameter of not less than 0.05 mum. The composition can be calcined at low temperatures to form a dielectric layer with high dimensional precision, said layer having a high dielectric constant and a low dielectric loss. Also provided are a dielectric and an electronic part prepared from the composition.

The power module of the invention includes: a cell stack including a plurality of cells that are stacked; a pair of end plates pressing and sandwiching the cell stack in the stacking direction; and a binding member joining the pair of end plates. The binding member has a recess that fits with an edge of the cell stack extending in the stacking direction. This ensures that the cells are bound together firmly and reliably.

Disclosed is a photosensitive resin composition for forming a laser engravable printing element, comprising: (a) 100 parts by weight of a resin which is a plastomer at 20° C., wherein the resin has a number average molecular weight (Mn) of from 1,000 to 100,000 and has a polymerizable unsaturated group in an amount such that the average number of the polymerizable unsaturated group per molecule is at least 0.7, (b) 5 to 200 parts by weight, relative to 100 parts by weight of resin (a), of an organic compound having an Mn of less than 1,000 and having at least one polymerizable unsaturated group per molecule, and (c) 1 to 100 parts by weight, relative to 100 parts by weight of resin (a), of an inorganic porous material. Also disclosed is a laser engravable printing element formed from the above-mentioned resin composition.

A fuel cell battery is operated with an afterburning at the periphery of a cell stack. Each cell of the battery has at least one entry point for air or another gas containing oxygen. The afterburning is provided inside a ring-shaped space about the cell stack. The named entry points are communicatingly connected as a totality or in each case group-wise via at least one air space, which extends axially along the cell stack and which stands in direct contact with the latter. Each air space is separated from an afterburner chamber, which likewise forms a space which communicates axially along the cell stack, by at least one wall.

Methods and systems are disclosed for autonomously building a predictive model of outcomes. A most-predictive set of signals Sk is identified out of a set of signals s1, s2, . . . , sD for each of one or more outcomes ok. A set of probabilistic predictive models ôk=Mk(Sk) is autonomously learned, where ôk is a prediction of outcome ok derived from the model Mk that uses as inputs values obtained from the set of signals Sk. The step of autonomously learning is repeated incrementally from data that contains examples of values of signals s1, s2, . . . , sD and corresponding outcomes o1, o2, . . . , oK. Various embodiments are also disclosed that apply predictive models to various physiological events and to autonomous robotic navigation.

An engine block die-casting apparatus of the present invention includes a stationary element, an ejector holder block adapted to be operatively movable to and from the stationary element, and an ejector box. The apparatus also includes a pair of side slide cores and at least one bank core slide assembly that is slidably mounted and mechanically actuated within the ejector holder block. The stationary element, the ejector holder block, the pair of side slide cores, and the bank core slide assembly are adapted to be moved proximate each other so as to create a closed die-cast cavity and to be drawn apart from one another to allow extraction of the cast engine block.

A plastic ferrule including a centering section having a tubular body. The tubular body is provided with an outer-circumferential centering reference surface and a center axis of the centering reference surface. The centering section includes a resinous outer tube part having the outer-circumferential centering reference surface and provided, along the center axis, with a coated-fiber holding bore for securely accommodating a coated optical fiber and a fixing bore extending from the coated-fiber holding bore, the outer tube part forming a tubular wall with a generally uniform thickness as a whole. The centering section also includes an inner tube part fixed to the fixing bore of the outer tube part and provided along the center axis with an uncoated-fiber holding bore, communicating with the coated-fiber holding bore, for securely accommodating an uncoated optical fiber.

The method according to the present invention comprises: a first curving step of curving a plate-like metallic blank (1) such that the blank (1) is convexed; a bending step of bending, with the outer peripheral edge portion (1e) of the blank (1) restrained from radially outwardly extending, the resulting arcuate portion (1b) of the curved blank (1) in the direction opposite to the convex direction thereof, so that a boss (6) is formed; and a second curving step of pushing the inner peripheral portion of a flat portion (5) of the blank (1) in the direction opposite to the boss projecting direction. Through the steps above-mentioned, the blank (1) is provided at the center thereof with the case-like boss (6) projecting in one direction from one lateral side of the blank (1) and at the outer periphery thereof with a peripheral wall (7) projecting in the same direction in which the boss (6) projects. Thus, with a small press machine, there can be securely formed, with high precision, a boss having a desired diameter and a desired projecting height, without the original thickness of the blank (1) decreased so much in the course of production steps.

An internal combustion engine includes a cylinder block formed of a metal material, a cylinder liner formed of a metal material different from that of the cylinder block, the cylinder liner being fitted onto the cylinder block, a water jacket provided between the cylinder block and the cylinder liner for retaining a coolant, and a sealing member provided in contact with the cylinder block and the cylinder liner, the sealing member preventing leakage of the coolant from the water jacket. The cylinder block and the cylinder liner have respective positioning surfaces, each of which determines a relative position against the other. The positioning surfaces are provided on the opposite side of the sealing member from the water jacket. The cylinder block and the cylinder liner are spaced apart from each other, in between the sealing member and the water jacket.

A fluorine-containing cyclic olefin polymer composition of the present invention includes a fluorine-containing cyclic olefin polymer (A) containing a repeating structural unit represented by the general formula (1) and having a fluorine atom content rate of 40 to 75% by mass; a photocurable compound (B); and a photocuring initiator (C).

A method of processing a target object includes (a) exposing a resist mask to active species of hydrogen generated by exciting plasma of a hydrogen-containing gas within a processing vessel while the target object is mounted on a mounting table provided in the processing vessel; and (b) etching a hard mask layer by exciting plasma of an etchant gas within the processing vessel after the exposing of the resist mask to the active species of hydrogen. The plasma is excited by applying of a high frequency power for plasma excitation to an upper electrode. In the method, a distance between the upper electrode and the mounting table in the etching of the hard mask layer ((b) process) is set to be larger than a distance between the upper electrode and the mounting table in the exposing of the resist mask to the active species of hydrogen ((a) process).

The invention provides a transfer technique by which the dimensional precision of a thin-film device is not deteriorated, even if the device is produced by transferring a fine structure or a thin-film circuit layer onto a substrate with an inferior shape-stability. The method includes: forming a fine structure or a thin-film circuit layer on a first substrate using a photolithographic patterning process; shifting the fine structure or the thin-film circuit layer from the first substrate onto a second substrate, or shifting the fine structure or the thin-film circuit layer from the first substrate onto the second substrate via a third substrate; and forming a thin-film pattern on the fine structure or the thin-film circuit layer shifted onto the second substrate by a non-photolithographic method.

The present invention aims to provide a composition for a holographic recording medium which allows for producing a disc medium filled with a photopolymer by a very simple method such as hot press method, with high dimensional precision, high uniformity and high productivity. The composition for a holographic recording medium contains a binder, a polymerizable polymer, and a photopolymerization initiator, wherein the binder is a polymer capable of forming a three-dimensional crosslinking structure by non-polymerization reaction. An aspect is preferable in which the polymer capable of forming a three-dimensional crosslinking structure by non-polymerization reaction is at least one of (A) a polymer which has an ionic group and formed a salt with the counterion of the ionic group and (B) a polymer having a crystal structure.

A putter head includes a head main body that defines a periphery of the putter head, and a bar that crosses a void space of the head main body in a forward-backward direction. The bar is inserted into insertion holes and fixed thereto by an adhesive. A face member is mounted on a recess. A shaft fixing hole is provided to pass through the bar from the top surface of a front portion of the head main body. A shaft is inserted into the shaft fixing hole and fixed thereto by an adhesive.

A method for data-driven anomaly detection may include monitoring a plurality of parameters associated with a plurality of subsystems of a system. The method may also include collecting data corresponding to each of the plurality of parameters from the plurality of subsystems and generating monitoring quantities based on the data. The method may also include determining if any quantities in the monitoring quantities exceed a predetermined limit. A contribution plot may be generated corresponding to each of the parameters in response to any of the quantities exceeding the predetermined limit. The method may further include determining which parameter is likely to cause an effect based on the contribution plot.