摘要:
A structured prediction model learning apparatus, method, program, and recording medium maintain prediction performance with a smaller amount of memory. An auxiliary model is introduced by defining the auxiliary model parameter set θ(k) with a log-linear model. A set Θ of auxiliary model parameter sets which minimizes the Bregman divergence between the auxiliary model and a reference function indicating the degree of pseudo accuracy is estimated by using unsupervised data. A base-model parameter set λ which minimizes an empirical risk function defined beforehand is estimated by using supervised data and the set Θ of auxiliary model parameter sets.
摘要:
A semiconductor device includes a semiconductor substrate; an n-channel MOS transistor including a first gate insulating film provided on a p-type layer, a first gate electrode made of TiN, and a first upper gate electrode made of semiconductor doped with impurities; and a p-channel MOS transistor including a second gate insulating film provided on an n-type layer, a second gate electrode including at least as a part, a TiN layer made of TiN crystal in which a ratio of (111) orientation/(200) orientation is about 1.5 or more, and a second upper gate electrode made of semiconductor doped with impurities.
摘要:
A blade for wind power generation includes a lightning receptor configured to form a part of the surface of the blade, and a lightning receiving protrusion that protrudes outward from the surface of the lightning receptor. The lightning receiving protrusion is provided to a surface boundary between the blade and the lightning receptor. The surface boundary configured to protect against the lightning is positioned within a radius of a circle, which is centered on a tip of the lightning receiving protrusion and which has a radius of twice the length of the lightning receiving protrusion. A point where electric field concentration on the surface of the lightning receptor is greatest is moved from the boundary to the tip of the lightning receiving protrusion, so as to significantly reduce a possibility of blade damage associated with lightning striking the boundary when being stroke by the lightning.
摘要:
The present invention provides a reflection film, a reflection film laminate which are less likely to undergo agglomeration or sulfidation of an Ag thin film due to heat, and a LED, an organic EL display, and an organic EL illuminating instrument, each including any of these. The reflection film in accordance with the present invention is a reflection film formed on a substrate, characterized by being an Ag alloy film including Ag as a main component, and Bi in an amount of 0.02 atomic percent or more, and further including one or more of V, Ge, and Zn in a total content of 0.02 atomic percent or more, and satisfying the following expression (1): 7×[A]+13×[Bi]≦8 (1) where [A] (atomic percent) denotes the content of one or more of the V, Ge, and Zn, and [Bi] (atomic percent) denotes the content of Bi. Whereas, a reflection film laminate in accordance with the present invention is a reflection film laminate formed on a substrate, characterized by including: a first film comprising an Ag alloy film including Ag as a main component, and Bi in an amount of 0.02 atomic percent or more, and further including one or more of V, Ge, and Zn in a total content of 0.02 atomic percent or more, and satisfying the expression (1); and a second film including a Si oxide formed on the first film.
摘要:
Any packet loss is detected very quickly by means of only a series of sequence number in a multi-path environment where a transmitter and a receiver are connected to each other by way of a plurality of networks when no inversion of sequence arises in any of the networks. A communication apparatus includes a plurality of sequence buffers arranged at each network to accumulate packets until a sequence acknowledgement and an absence detecting section adapted to determine the occurrence of an absence of a packet when one or more packets are accumulated in all the sequence buffers. With this arrangement, the absence detecting section of the receiver monitors the packets staying in the sequence guaranteeing buffer arranged in each of the network, paying attention to the characteristic that packets are stored in the sequence buffers of all the networks when a packet loss takes place.
摘要:
A driving apparatus (100) is provided with: a base part (110); a driven part (130) capable of rotating around an axis which is along one direction (Y-axis); an elastic part (120) which connects the base part and the driven part and which extends along the one direction; and an applying part (140) for applying a force for expanding and contracting the elastic part along the one direction as an excitation force for rotating the driven part such that the driven part resonates at a resonance frequency determined by the elastic part and the driven part.
摘要:
A structured prediction model learning apparatus, method, program, and recording medium maintain prediction performance with a smaller amount of memory. An auxiliary model is introduced by defining the auxiliary model parameter set θ(k) with a log-linear model. A set Θ of auxiliary model parameter sets which minimizes the Bregman divergence between the auxiliary model and a reference function indicating the degree of pseudo accuracy is estimated by using unsupervised data. A base-model parameter set λ which minimizes an empirical risk function defined beforehand is estimated by using supervised data and the set Θ of auxiliary model parameter sets.
摘要:
A packet communication system includes a virtual machine unit; an origin-of-movement communication unit; a movement-destination communication unit; and a storage unit which stores a packet destined for the virtual machine unit while the virtual machine unit moves from the origin-of-movement communication unit to the movement-destination communication unit and transfers the stored packet to the movement-destination communication unit.
摘要:
A driving apparatus (100) is provided with: a base portion (110); a stage portion (130) on which a driven object (12) is mounted and which can be displaced; an elastic portion (120) which connects the base portion and the stage portion and which has elasticity to displace the stage portion in one direction; a first applying device (141, 142, 143, 22) for applying an excitation force for displacing the stage portion such that the stage portion is resonated in the one direction (Y axis) at a resonance frequency determined by the stage portion and the elastic portion; and a second applying device (161, 162, 22) for applying a driving force for displacing, in a stepwise manner or in a continuous manner, the stage portion or the driven object mounted on the stage in other direction (X axis).
摘要:
There are provided are a plurality of memory mats, a sub-word driver that accesses a normal memory cell irrespective of whether a row address to which access is requested is a defective address, a sub-word driver that accesses a redundant memory cell belonging to a memory mat different from the normal memory cell indicated by the row address, when the row address is a defective address. According to the present invention, the normal memory cell and a redundant memory cell belong to memory mats different to each other, and thus the normal memory cell can be accessed concurrently with determining operation of the repair determining circuit.