Abstract: A method for processing a voice command using a statistical dialog model determines a belief state as a probability distribution over states organized in a hierarchy with a parent-child relationship of nodes representing the states. The belief state includes the hierarchy of state variables defining probabilities of each state to correspond to the voice command and a probability of a state of a child node in the hierarchy is conditioned on a probability of a state of a corresponding parent node. A system action is selected based on the belief state.

Abstract: Thermal resistance is reduced from an element surface of a semiconductor chip to the rear surface of a semiconductor package. Split patterning of a metal is easily carried out, stress produced by a thermal expansion coefficient between silicon and metal is significantly reduced and environment reliability is improved. Low cost is realized by manufacturing a semiconductor package without using a TIM material. A semiconductor package is provided including a semiconductor chip including a first surface and a second surface opposed to the first surface and covered with a resin, an electrode being arranged over the first surface, a first wiring connected to the first surface directly or via a first opening arranged in the resin, and a second wiring connected to the second surface via a second opening arranged in the resin.

Abstract: The present invention provides an anti-influenza virus agent that targets biomolecules of host cells including human cells and a method of screening a candidate molecule for the anti-influenza virus agent. That is, the present invention is an anti-influenza virus agent that has an effect of suppressing expression or a function of a gene that encodes a protein having an effect of suppressing incorporation of an influenza virus vRNA or an NP protein into influenza virus-like particles in host cells and the gene is at least one selected from the group including JAK1 gene and the like.

Abstract: A method of setting a laser-light intensity value includes: emitting laser light, the laser light being excitation light, a fluorescent-dyed biological sample being irradiated with the excitation light and emitting light; detecting fluorescence emitted by the biological sample, and outputting a signal corresponding to a brightness value; prestoring relation information, the relation information including the plurality of laser-light intensity values, and information on at least one possible correlation between a phototoxicity degree and the brightness value in relation to each of the laser-light intensity values, the phototoxicity to the biological sample resulting from the laser light; generating a fluorescence image having the brightness value based on the output signal; calculating a brightness value representative of a ROI area based on the generated fluorescence image; and referring to the relation information, and determining a laser-light intensity value satisfying tolerance of the phototoxicity based

Abstract: A method and for training a language model to reduce recognition errors, wherein the language model is a recurrent neural network language model (RNNLM) by first acquiring training samples. An automatic speech recognition system (ASR) is appled to the training samples to produce recognized words and probabilites of the recognized words, and an N-best list is selected from the recognized words based on the probabilities. determining word erros using reference data for hypotheses in the N-best list. The hypotheses are rescored using the RNNLM. Then, we determine gradients for the hypotheses using the word errors and gradients for words in the hypotheses. Lastly, parameters of the RNNLM are updated using a sum of the gradients.

Abstract: [Object] It is an object to provide an injection method for injecting an electrolyte and an electrolyte injection apparatus which allow the electrolyte to be injected and filled into an electrode assembly within an outer can with favorable permeation, thereby easily manufacturing an electrolyte secondary battery having favorable cycle characteristics at good yield. [Solution] A solution injection nozzle 10 is inserted into a solution injection hole 101 of an outer can 100 in which an electrode assembly 110 is stored, and the solution injection hole 101 is hermetically sealed by a pressure reducing pad 11 provided so as to surround a periphery of the solution injection nozzle 10. An inside of the outer can 100 is made into a negative pressure through the pressure reducing pad 11, and an electrolyte L is supplied from the solution injection nozzle 10 into the outer can 100. The outer can 100 is rotated with the solution injection nozzle 10 as a rotation center.

Abstract: A method distinguishes components of a signal by processing the signal to estimate a set of analysis features, wherein each analysis feature defines an element of the signal and has feature values that represent parts of the signal, processing the signal to estimate input features of the signal, and processing the input features using a deep neural network to assign an associative descriptor to each element of the signal, wherein a degree of similarity between the associative descriptors of different elements is related to a degree to which the parts of the signal represented by the elements belong to a single component of the signal. The similarities between associative descriptors are processed to estimate correspondences between the elements of the signal and the components in the signal. Then, the signal is processed using the correspondences to distinguish component parts of the signal.

Abstract: A method estimates source signals from a mixture of source signals by first training an analysis model and a reconstruction model using training data. The analysis model is applied to the mixture of source signals to obtain an analysis representation of the mixture of source signals, and the reconstruction model is applied to the analysis representation to obtain an estimate of the source signals, wherein the analysis model utilizes an analysis linear basis representation, and the reconstruction model utilizes a reconstruction linear basis representation.

Abstract: A method and system processes utterances that are acquired either from an automatic speech recognition (ASR) system or text. The utterances have associated identities of each party, such as role A utterances and role B utterances. The information corresponding to utterances, such as word sequence and identity, are converted to features. Each feature is received in an input layer of a neural network(NN). A dimensionality of each feature is reduced, in a projection layer of the NN, to produce a reduced dimensional feature. The reduced dimensional feature is processed to provide probabilities of labels for the utterances.

Abstract: A stacked semiconductor package includes a first semiconductor package including a first circuit board and a first semiconductor device mounted on the first circuit board; a second semiconductor package including a second circuit board and a second semiconductor device mounted on the second circuit board, the second semiconductor package being stacked on the first semiconductor package; and a heat transfer member provided on the first semiconductor device and a part of the first circuit board, the part being around the first semiconductor device.

Abstract: The invention provides methods to identify agents useful to prevent, inhibit or treat viral infections, e.g., filovirus infections, as well as compositions having one or more agents to prevent, inhibit or treat viral infection.

Abstract: A spoken language understanding (SLU) system receives a sequence of words corresponding to one or more spoken utterances of a user, which is passed through a spoken language understanding module to produce a sequence of intentions. The sequence of words are passed through a first subnetwork of a multi-scale recurrent neural network (MSRNN), and the sequence of intentions are passed through a second subnetwork of the multi-scale recurrent neural network (MSRNN). Then, the outputs of the first subnetwork and the second subnetwork are combined to predict a goal of the user.

Abstract: A method processes an acoustic signal that is a mixture of a target signal and interfering signals by first enhancing the acoustic signal by a set of enhancement procedures to produce a set of initial enhanced signals. Then, an ensemble learning procedure is applied to the acoustic signal and the set of initial enhancement signals to produce features of the acoustic signal.

Abstract: A spoken language understanding (SLU) system receives a sequence of words corresponding to one or more spoken utterances of a user, which is passed through a spoken language understanding module to produce a sequence of intentions. The sequence of words are passed through a first subnetwork of a multi-scale recurrent neural network (MSRNN), and the sequence of intentions are passed through a second subnetwork of the multi-scale recurrent neural network (MSRNN). Then, the outputs of the first subnetwork and the second subnetwork are combined to predict a goal of the user.

Abstract: In a semiconductor device according to an embodiment, a second semiconductor chip is mounted on a first rear surface of a first semiconductor chip. Also, the first rear surface of the first semiconductor chip includes a first region in which a plurality of first rear electrodes electrically connected to the second semiconductor chip via a protrusion electrode are formed and a second region which is located on a peripheral side relative to the first region and in which a first metal pattern is formed. In addition, a protrusion height of the first metal pattern with respect to the first rear surface is smaller than a protrusion height of each of the plurality of first rear electrodes with respect to the first rear surface.

Abstract: A magnetic shielding package of a non-volatile magnetic memory element, including: a soft magnetic material support plate 12; a first insulating material layer 13 formed on the support plate; a non-volatile magnetic memory element 11 fixed on the first insulating material layer; a second insulating material layer 14 that encapsulates the memory element and the periphery thereof; in the second insulating material layer, a wiring layer 15, a soft magnetic layer 15b or 25 and a conductive portion 16 connecting an electrode of the circuit surface of the memory element and the wiring layer; and a magnetic shield part 17 containing a soft magnetic material arranged like a wall at a distance from a side surface of the memory element so as to surround the memory element side surface partially or entirely, the magnetic shield part being magnetically connected to the soft magnetic layer.

Abstract: A method distinguishes components of a signal by processing the signal to estimate a set of analysis features, wherein each analysis feature defines an element of the signal and has feature values that represent parts of the signal, processing the signal to estimate input features of the signal, and processing the input features using a deep neural network to assign an associative descriptor to each element of the signal, wherein a degree of similarity between the associative descriptors of different elements is related to a degree to which the parts of the signal represented by the elements belong to a single component of the signal. The similarities between associative descriptors are processed to estimate correspondences between the elements of the signal and the components in the signal. Then, the signal is processed using the correspondences to distinguish component parts of the signal.

Abstract: A program that may cause an information processing apparatus that executes a game to execute: storing, in a storage unit, a first parameter associated with a first game content arranged in virtual space and a second parameter associated with a second game content arranged in the virtual space; starting a match between the first game content and the second game content; automatically changing a third parameter regarding a positional relationship between the first game content and the second game content in the virtual space to be closer to a first target value at a certain time change rate; changing the second parameter associated with the second game content, on the basis of the third parameter and the first parameter associated with the first game content; and determining that a user wins the match when the second parameter becomes a second target value.

Abstract: Techniques for image stabilization may include detecting motion of an apparatus configured to display image data, the image data comprising one or more frames, a first frame of the one or more frames comprising a plurality of layers. The plurality of layers may be processed to correct for the detected motion. The processing may comprise applying a different degree of motion correction to a first layer of the plurality of layers than to a second layer of the plurality of layers. Such techniques may be performed via an apparatus comprising a display control unit configured to cause the image data to be displayed, and a motion correction unit configured to perform the processing.