Abstract: Deep Neural Network (DNN) training technique embodiments are presented that train a DNN while exploiting the sparseness of non-zero hidden layer interconnection weight values. Generally, a fully connected DNN is initially trained by sweeping through a full training set a number of times. Then, for the most part, only the interconnections whose weight magnitudes exceed a minimum weight threshold are considered in further training. This minimum weight threshold can be established as a value that results in only a prescribed maximum number of interconnections being considered when setting interconnection weight values via an error back-propagation procedure during the training. It is noted that the continued DNN training tends to converge much faster than the initial training.

Abstract: Provided herein are products, uses and method classifying a subject afflicted with breast cancer according to a ClinicoMolecular Triad Classification (CMTC)-1, CMTC-2 or CMTC-3 class. The method involves: (i) determining a subject expression profile, said subject expression profile comprising the mRNA expression levels of a plurality of genes that classify breast cancer into three groups by hierarchal clustering TN and Her2+ breast cancers into one class (CMTC genes), in a breast cancer cell sample taken from said subject; (ii) calculating a measure of similarity between said subject expression profile, and one or more of: a) a CMTC-1 reference profile, b) a CMTC-2 reference profile, and c) a CMTC-3 reference profile; and (iii) classifying said subject.

Abstract: A deep tensor neural network (DTNN) is described herein, wherein the DTNN is suitable for employment in a computer-implemented recognition/classification system. Hidden layers in the DTNN comprise at least one projection layer, which includes a first subspace of hidden units and a second subspace of hidden units. The first subspace of hidden units receives a first nonlinear projection of input data to a projection layer and generates the first set of output data based at least in part thereon, and the second subspace of hidden units receives a second nonlinear projection of the input data to the projection layer and generates the second set of output data based at least in part thereon. A tensor layer, which can converted into a conventional layer of a DNN, generates the third set of output data based upon the first set of output data and the second set of output data.

Abstract: A method for providing position information and identifying positions of others, using an electronic apparatus, includes detecting first position information regarding a current position of the electronic apparatus, transmitting a position inquiry request and information regarding a predetermined time period to an external apparatus, receiving second position information regarding a position of the external apparatus from the external apparatus for the predetermined time period, and displaying positions of the electronic apparatus and the external apparatus on a screen using the first position information and the second position information.

Abstract: A content providing method of an electronic apparatus communicably connected with an external apparatus is provided. The method includes receiving a command for transferring content to the external apparatus; checking whether the content includes information indicating that the content is transferable to the external apparatus, the information being set by determining, based on a meta data included in the content, whether a predetermined condition is satisfied; and transferring the content, based on the checked information, to the external apparatus.

Abstract: A saturation adjusting apparatus for processing a pixel, which has three color components each having a value falling within a range defined by upper and lower extreme values, of a RGB color model includes an extreme value controller and a component adjuster. The extreme value controller determines maximum and minimum extreme value thresholds for ensuring that the values of the color components of the pixel after undergoing linear color correction processing based on a correction indicator fall within the range defined by the upper and lower extreme values. The component adjuster includes a decision-making unit for choosing the correction indicator from a group of values which includes the maximum extreme value threshold, the minimum extreme value threshold, and a saturation setting. The component adjuster further includes a color corrector for performing linear color correction processing on the three color components of the pixel using the correction indicator.

Abstract: A control device which controls the power of a plurality of apparatuses in a home network, and a method thereof are provided. The control device includes an input unit which receives a power off command, and a control unit which concurrently turns off apparatuses which are currently turned on among the plurality of apparatuses in response to the power off command received by the input unit. Therefore, it is possible for a user to concurrently turn off currently running apparatuses so user convenience can increase.

Abstract: A method of generating interaction activity information including connecting a second apparatus to a first apparatus capable of communicating with each other, receiving first information about a function that is capable of being performed in the first apparatus, from the first apparatus, and generating second information about a function that is used interactively between the first apparatus and the second apparatus, based on the first information.

Abstract: An electronic apparatus controls a peripheral device by using a relay apparatus. The electronic apparatus includes an interface connected to the peripheral device; a communication unit that performs communication with the relay apparatus; a receiver that receives a control signal for controlling the electronic apparatus; and a controller that, when the control signal is received, performs an operation based on the control signal, and controls the communication unit to transmit information about the performed operation and information about the peripheral device to the relay apparatus in order to control the peripheral device to perform an operation that corresponds to the performed operation.

Abstract: A content providing method is provided involving an electronic apparatus and an external apparatus communicably connected with each other. A play command of the external apparatus for a particular content is received. at the electronic apparatus When the particular content is set to a content transferable to the external apparatus, the electronic apparatus sends the particular content to the external apparatus in response to the play command. Thus, a user can check only specified photos in the external apparatus.

Abstract: Disclosed is a control method for a display apparatus and a mobile terminal which includes a camera for taking a photograph and generating an image; a voice input unit which is used to input a voice; a user input unit which receives a user's input; a display unit which displays the image thereon; a wireless communication unit which communicates with a counterpart mobile terminal through a wireless network; and a controller which performs a video call by transmitting to the counterpart mobile terminal video call data comprising a video generated by the camera and a voice input by the voice input unit according user input, displays on the display unit a shared image during the video call, and transmits to the counterpart mobile terminal the shared image added to the video call data.

Abstract: A method is disclosed herein that includes an act of causing a processor to access a deep-structured, layered or hierarchical model, called a deep convex network, retained in a computer-readable medium, wherein the deep-structured model comprises a plurality of layers with weights assigned thereto. This layered model can produce the output serving as the scores to combine with transition probabilities between states in a hidden Markov model and language model scores to form a full speech recognizer. Batch-based, convex optimization is performed to learn a portion of the deep convex network's weights, rendering it appropriate for parallel computation to accomplish the training. The method can further include the act of jointly substantially optimizing the weights, the transition probabilities, and the language model scores of the deep-structured model using the optimization criterion based on a sequence rather than a set of unrelated frames.

Abstract: A method of providing an image of an image capture apparatus including receiving a first input instructing the image capture apparatus to photograph a sequence of images, photographing the sequence of images, in response to receiving the first input, storing, by a control unit of the image capture apparatus, the sequence of images in a memory of the image capture apparatus, receiving a second input, and providing a plurality of images from among the sequence of images, in response to receiving the second input.

Abstract: The invention provides novel immune regulatory oligonucleotides (IRO) as antagonist of TLRs and methods of use thereof. These IROs have unique sequences that inhibit or suppress TLR-mediated signaling in response to a TLR ligand or TLR agonist. The methods may have use in the prevention and treatment of cancer, an autoimmune disorder, airway inflammation, inflammatory disorders, infectious disease, skin disorders, allergy, asthma or a disease caused by a pathogen.

Abstract: A method of providing an image of an image capture apparatus, the method including executing an application for a remote device to remotely control the image capture apparatus to successively capture a sequence of video frames, receiving from the remote device a first input instructing the image capture apparatus to successively capture the sequence of video frames, successively capturing the sequence of video frames and streaming the sequence of video frames to the remote device, in response to receiving the first input, receiving from the remote device a second input instructing the image capture apparatus to capture a still image, and providing the still image, in response to receiving the second input.

Abstract: A tensor deep stacked neural (T-DSN) network for obtaining predictions for discriminative modeling problems. The T-DSN network and method use bilinear modeling with a tensor representation to map a hidden layer to the predication layer. The T-DSN network is constructed by stacking blocks of a single hidden layer tensor neural network (SHLTNN) on top of each other. The single hidden layer for each block then is separated or divided into a plurality of two or more sections. In some embodiments, the hidden layer is separated into a first hidden layer section and a second hidden layer section. These multiple sections of the hidden layer are combined using a product operator to obtain an implicit hidden layer having a single section. In some embodiments the product operator is a Khatri-Rao product. A prediction is made using the implicit hidden layer and weights, and the output prediction layer is consequently obtained.

Abstract: The invention relates to the therapeutic use of stabilized oligoribonucleotides as immune modulatory agents for immune therapy applications. Specifically, the invention provides RNA based oligoribonucleotides with improved nuclease and RNase stability and that have immune modulatory activity through TLR7 and/or TLR8.

Abstract: A method is disclosed herein that includes an act of causing a processor to access a deep-structured, layered or hierarchical model, called deep convex network, retained in a computer-readable medium, wherein the deep-structured model comprises a plurality of layers with weights assigned thereto. This layered model can produce the output serving as the scores to combine with transition probabilities between states in a hidden Markov model and language model scores to form a full speech recognizer. The method makes joint use of nonlinear random projections and RBM weights, and it stacks a lower module's output with the raw data to establish its immediately higher module. Batch-based, convex optimization is performed to learn a portion of the deep convex network's weights, rendering it appropriate for parallel computation to accomplish the training.

Abstract: The invention relates to the therapeutic use of oligonucleotides or oligonucleotide analogs as immunostimulatory agents in immunotherapy applications. The invention provides methods for enhancing the immune response caused by immunostimulatory oligonucleotide compounds.

Abstract: Described is a technology by which a deep-structured (multiple layered) conditional random field model is trained and used for classification of sequential data. Sequential data is processed at each layer, from the lowest layer to a final (highest) layer, to output data in the form of conditional probabilities of classes given the sequential input data. Each higher layer inputs the conditional probability data and the sequential data jointly to output further probability data, and so forth, until the final layer which outputs the classification data. Also described is layer-by-layer training, supervised or unsupervised. Unsupervised training may process raw features to minimize average frame-level conditional entropy while maximizing state occupation entropy, or to minimize reconstruction error.