A method and apparatus of training a neural network. The method and apparatus include creating a model for a desired function as a multi-dimensional function, determining if the created model fits a simple finite geometry model, and generating a Radon transform to fit the simple finite geometry model. The desired function is fed through the Radon transform to generate weights. A multilayer perceptron of the neural network is trained using the weights.

An occurrence description scheme that describes an occurrence of a semantic entity in multimedia content is encoded into a content description for the content. The occurrence description scheme is extracted from the content description and used by an application to search, filter or browse the content when a full structural or semantic description of the content is not required.

The present invention provides improved techniques for spatial representation of data and browsing based on similarity. For example, improved techniques for spatial representation of image data and browsing the image data based on the similarities (or dissimilarities) of the images are provided. In one embodiment, a hierarchical MultiDimensional Scaling (MDS) database for a set of images is provided, which allows for computationally efficient querying and updating of an image database. In one em...

A method and apparatus for comparing data is described. In one embodiment, an exemplary method includes receiving a first set of data pertaining a first object and a second set of data pertaining to a second object, and comparing the first object with the second object using an earth mover's distance method that is based on computation of a series of Hausdorff distances.

The invention relates to a brush comprising a plurality of bristles and each bristle comprises a plurality of fibers. A palette of virtual paint is provided for the brush to contact at least one paint. The brush contacts the tablet and applies paint to the surface of the tablet or image that is created.

A method and apparatus for determining quality of a description are described. According to one embodiment, an exemplary method for determining quality of a description includes posing a classification task concerning at least one audiovisual object to a descriptive method that is used to create the description, generating a set of probabilities from a result of the classification task, and measuring an entropy of the result using the set of probabilities.

The present invention provides improved techniques for spatial representation of data and browsing based on similarity. For example, improved techniques for spatial representation of image data and browsing the image data based on the similarities (or dissimilarities) of the images are provided. In one embodiment, a hierarchical MultiDimensional Scaling (MDS) database for a set of images is provided, which allows for computationally efficient querying and updating of an image database. In one em...

A method and an apparatus of compressing data. The method and apparatus include constructing a neural network having a specific geometry using a finite and discrete Radon transform. The data is then fed through the neural network to produce a transformed data stream. The transformed data stream is thresholded. A fixed input signal is fed back through the neural network to generate a decoding calculation of an average value. The thresholded data stream is entropy encoded.

A method, apparatus and article of manufacture for updating a new node in a multidimensional scaling (MDS) database having an existing node. The new node is compared to the existing node to obtain a disparity value. A distance value is calculated between the new node and the existing node, a sum of differences value is calculated for the disparity value and the distance value, and a sum of squares value is calculated for the disparity value and the distance value. The position of the new node is...

A method and an apparatus of designing a set of wavelet basis trained to fit a particular problem. The method and apparatus include constructing a neural network of arbitrary complexity using a discrete and finite Radon transform, feeding an input wavelet prototype through the neural network and its backpropagation to produce an output, and modifying the input wavelet prototype using the output.