Abstract: A method of modelling data, comprising: training an objective function of a linear classifier, based on a set of labeled data, to derive a set of classifier weights; defining a posterior probability distribution on the set of classifier weights of the linear classifier; approximating a marginalized loss function for an autoencoder as a Bregman divergence, based on the posterior probability distribution on the set of classifier weights learned from the linear classifier; and classifying unlabeled data using the autoencoder according to the marginalized loss function.

Abstract: A halogen-free modified high-filling recyclable plastic board is provided in this disclosure, which includes a substrate layer and a printed layer and a protective layer disposed sequentially on the substrate layer from bottom to top. Raw materials of the substrate layer include, by weight in percent, 20 to 25% of PEAT resin, 70 to 75% of stone powder, 0.5 to 0.8% of chain extender, 1 to 2% of white mineral oil, 3 to 6% PE, and 0.4 to 0.8% stearic acid. The plastic board according to the present disclosure is formed using a hot press process, without glue bonding and with good integrity; and the manufactured board is large in surface tension, its surface is easy to be processed and a substrate layer thereof has good compatibility with a printed layer and a protective layer, which can be recycled as a whole.

Abstract: A plastic floor board processing technology using digital printing, aiming to solve the problem relating to the high production cost, comprising the steps of: preparing a base material; blending the base material; extruding the blended base material into a mold to form a stone-plastic base material; adjusting a gap between a surface embossing roll and a bottom embossing roll to enable the stone-plastic base material to pass through the gap; generating embossing patterns and positioning marks at equal intervals on a surface of the stone-plastic base material; cooling the stone-plastic base material; cutting the stone-plastic base material into plastic floorboards; using a digital printer to print the plastic floorboards. According to the present disclosure, patterns are directly printed on the surface of the stone-plastic base material, which avoids the processes of arranging a color film and a wear layer, lowers the production cost and improves the production efficiency.

Abstract: A manufacturing method of an ultra-matte board, which includes a semi-curing step, an irradiating step with an ultraviolet excimer lamp and a full curing step. The semi-curing step includes: coating a UV paint on a substrate, and irradiating the coated substrate with an ultraviolet lamp with a wavelength ranging from 350 nm to 450 nm to semi-cure the UV paint; the irradiating step with the ultraviolet excimer lamp includes: placing the semi-cured substrate in an inert gas atmosphere and irradiating the substrate with an ultraviolet excimer lamp for 2 s to 20 s; and the full curing step includes: irradiating with an ultraviolet lamp with a wavelength ranging from 350 nm to 450 nm to fully cure the UV paint to obtain the ultra-matte board.

Abstract: A method of modelling data, comprising: training an objective function of a linear classifier, based on a set of labeled data, to derive a set of classifier weights; defining a posterior probability distribution on the set of classifier weights of the linear classifier; approximating a marginalized loss function for an autoencoder as a Bregman divergence, based on the posterior probability distribution on the set of classifier weights learned from the linear classifier; and classifying unlabeled data using the autoencoder according to the marginalized loss function.

Abstract: A method of modelling data, comprising: training an objective function of a linear classifier, based on a set of labeled data, to derive a set of classifier weights; defining a posterior probability distribution on the set of classifier weights of the linear classifier; approximating a marginalized loss function for an autoencoder as a Bregman divergence, based on the posterior probability distribution on the set of classifier weights learned from the linear classifier; and classifying unlabeled data using the autoencoder according to the marginalized loss function.

Abstract: A method for forming a stereoscopic pattern of a plastic floorboard, comprising: positioning a base board on a digital printing unit, wherein the base board has an outer surface; using the digital printing unit to print pigment on the outer surface of the base board according to a preset pattern to form a pattern layer, wherein the pattern of the pattern layer corresponds to or does not correspond to the stereoscopic pattern; using a first forming unit to form a protection layer on the pattern layer with a transparent melt plastic raw material; and using a second forming unit to form the stereoscopic pattern having a concave-convex structure on the protection layer.

Abstract: A plastic floor board processing technology using digital printing, aiming to solve the problem relating to the high production cost, comprising the steps of: preparing a base material; blending the base material; extruding the blended base material into a mold to form a stone-plastic base material; adjusting a gap between a surface embossing roll and a bottom embossing roll to enable the stone-plastic base material to pass through the gap; generating embossing patterns and positioning marks at equal intervals on a surface of the stone-plastic base material; cooling the stone-plastic base material; cutting the stone-plastic base material into plastic floorboards; using a digital printer to print the plastic floorboards. According to the present disclosure, patterns are directly printed on the surface of the stone-plastic base material, which avoids the processes of arranging a color film and a wear layer, lowers the production cost and improves the production efficiency.

Abstract: Systems and Methods for multi-modal or multimedia image retrieval are provided. Automatic image annotation is achieved based on a probabilistic semantic model in which visual features and textual words are connected via a hidden layer comprising the semantic concepts to be discovered, to explicitly exploit the synergy between the two modalities. The association of visual features and textual words is determined in a Bayesian framework to provide confidence of the association. A hidden concept layer which connects the visual feature(s) and the words is discovered by fitting a generative model to the training image and annotation words. An Expectation-Maximization (EM) based iterative learning procedure determines the conditional probabilities of the visual features and the textual words given a hidden concept class. Based on the discovered hidden concept layer and the corresponding conditional probabilities, the image annotation and the text-to-image retrieval are performed using the Bayesian framework.

Abstract: The invention discloses a chamfered plastic floor. The chamfered plastic floor is formed by splicing a plurality of substrates, wherein at least one side edge of an upper surface of each substrate is provided with a surface descending portion; the surface descending portion is disposed along a length direction of the edge of the substrate; adjacent substrates are spliced at the surface descending portions to form a boundary area, and a surface of the boundary area is covered with an antifouling layer. The invention provides a chamfered plastic floor which not only can present the boundaries between adjacent floors, but also facilitates the cleaning of the boundary areas between adjacent floors.

Abstract: Multimodal data mining in a multimedia database is addressed as a structured prediction problem, wherein mapping from input to the structured and interdependent output variables is learned.

Abstract: A method of modelling data, comprising: training an objective function of a linear classifier, based on a set of labeled data, to derive a set of classifier weights; defining a posterior probability distribution on the set of classifier weights of the linear classifier; approximating a marginalized loss function for an autoencoder as a Bregman divergence, based on the posterior probability distribution on the set of classifier weights learned from the linear classifier; and classifying unlabeled data using the autoencoder according to the marginalized loss function.

Abstract: Systems and Methods for multi-modal or multimedia image retrieval are provided. Automatic image annotation is achieved based on a probabilistic semantic model in which visual features and textual words are connected via a hidden layer comprising the semantic concepts to be discovered, to explicitly exploit the synergy between the two modalities. The association of visual features and textual words is determined in a Bayesian framework to provide confidence of the association. A hidden concept layer which connects the visual feature(s) and the words is discovered by fitting a generative model to the training image and annotation words. An Expectation-Maximization (EM) based iterative learning procedure determines the conditional probabilities of the visual features and the textual words given a hidden concept class. Based on the discovered hidden concept layer and the corresponding conditional probabilities, the image annotation and the text-to-image retrieval are performed using the Bayesian framework.

Abstract: Multimodal data mining in a multimedia database is addressed as a structured prediction problem, wherein mapping from input to the structured and interdependent output variables is learned.

Abstract: Multimodal data mining in a multimedia database is addressed as a structured prediction problem, wherein mapping from input to the structured and interdependent output variables is learned.

Abstract: The present invention features a qualitative method to detect independent motion revealed in successive frames of a compressed surveillance MPEG video stream using linear system consistency analysis without decompression of the stream, identifying the segments containing independent motion in a real-time or faster manner, for the retrieval of these segments. The linear system is constructed using the macroblocks of MPEG compressed video frames. The normal flow value of the macroblock is obtained by taking the dot product between the macroblock gradient vector, computed by averaging the four block gradient vectors, and the motion vector of this macroblock. The normal flow value is filtered for inclusion in the linear system, and the statistic of the matrices of the resulting linear system is determined, filtered to screen out false negatives and outliers, and used to determine the presence or absence of independent motion.

Abstract: A general model is provided which provides collective factorization on related matrices, for multi-type relational data clustering. The model is applicable to relational data with various structures. Under this model, a spectral relational clustering algorithm is provided to cluster multiple types of interrelated data objects simultaneously. The algorithm iteratively embeds each type of data objects into low dimensional spaces and benefits from the interactions among the hidden structures of different types of data objects.

Abstract: Relational clustering has attracted more and more attention due to its phenomenal impact in various important applications which involve multi-type interrelated data objects, such as Web mining, search marketing, bioinformatics, citation analysis, and epidemiology. A probabilistic model is presented for relational clustering, which also provides a principal framework to unify various important clustering tasks including traditional attributes-based clustering, semi-supervised clustering, co-clustering and graph clustering. The model seeks to identify cluster structures for each type of data objects and interaction patterns between different types of objects. Under this model, parametric hard and soft relational clustering algorithms are provided under a large number of exponential family distributions.

Abstract: Multimodal data mining in a multimedia database is addressed as a structured prediction problem, wherein mapping from input to the structured and interdependent output variables is learned.

Abstract: A general model is provided which provides collective factorization on related matrices, for multi-type relational data clustering. The model is applicable to relational data with various structures. Under this model, a spectral relational clustering algorithm is provided to cluster multiple types of interrelated data objects simultaneously. The algorithm iteratively embeds each type of data objects into low dimensional spaces and benefits from the interactions among the hidden structures of different types of data objects.