21554 results about "Feature vector" patented technology

A computer implemented method and system for gesture category recognition and training. Generally, a gesture is a hand or body initiated movement of a cursor directing device to outline a particular pattern in particular directions done in particular periods of time. The present invention allows a computer system to accept input data, originating from a user, in the form gesture data that are made using the cursor directing device. In one embodiment, a mouse device is used, but the present invention is equally well suited for use with other cursor directing devices (e.g., a track ball, a finger pad, an electronic stylus, etc.). In one embodiment, gesture data is accepted by pressing a key on the keyboard and then moving the mouse (with mouse button pressed) to trace out the gesture. Mouse position information and time stamps are recorded. The present invention then determines a multi-dimensional feature vector based on the gesture data. The feature vector is then passed through a gesture category recognition engine that, in one implementation, uses a radial basis function neural network to associate the feature vector to a pre-existing gesture category. Once identified, a set of user commands that are associated with the gesture category are applied to the computer system. The user commands can originate from an automatic process that extracts commands that are associated with the menu items of a particular application program. The present invention also allows user training so that user-defined gestures, and the computer commands associated therewith, can be programmed into the computer system.

A subscriber characterization system is presented in which the subscriber's selections are monitored, including monitoring of the time duration programming is watched, the volume at which the programming is listened to, and any available information regarding the type of programming, including category and sub-category of the programming. The characterization system can extract textual information related to the programming from closed captioning data, electronic program guides, or other text sources associated with the programming. The extracted information is used to form program characteristics vectors. The programming characteristics vectors can be used in combination with the subscriber selection data to form a subscriber profile. Heuristic rules indicating the relationships between programming choices and demographics can be applied to generate additional probabilistic information regarding demographics and programming and product interests.

The present invention provides methods and systems for patient-specific seizure onset detection. In one embodiment, at least one EEG waveform of the patient is recorded, and at least one epoch (sample) of the waveform is extracted. The waveform sample is decomposed into one or more subband signals via a wavelet decomposition of the waveform sample, and one or more feature vectors are computed based on the subband signals. A seizure onset can then be identified based on classification of the feature vectors to a seizure or a non-seizure class by comparing the feature vectors with a decision measure previously computed for that patient. The decision measure can be derived based on reference seizure and non-seizure EEG waveforms of the patient. In another aspect, similar methodology is employed for automatic detection of alpha waves. In other aspects, the invention provides diagnostic and imaging systems that incorporate the above seizure-onset and alpha-wave detection methodology.

A method for classifying or comparing objects includes detecting points of interest within two objects, computing feature descriptors at said points of interest, forming a multi-resolution histogram over feature descriptors for each object and computing a weighted intersection of multi-resolution histogram for each object. An alternative embodiment includes a method for matching objects by defining a plurality of bins for multi-resolution histograms having various levels and a plurality of cluster groups, each group having a center, for each point of interest, calculating a bin index, a bin count and a maximal distance to the bin center and providing a path vector indicative of the bins chosen at each level. Still another embodiment includes a method for matching objects comprising creating a set of feature vectors for each object of interest, mapping each set of feature vectors to a single high-dimensional vector to create an embedding vector and encoding each embedding vector with a binary hash string.

A method and system for training a user authentication by voice signal are described. In one embodiment, a set of feature vectors are decomposed into speaker-specific recognition units. The speaker-specific recognition units are used to compute distribution values to train the voice signal. In addition, spectral feature vectors are decomposed into speaker-specific characteristic units which are compared to the speaker-specific distribution values. If the speaker-specific characteristic units are within a threshold limit of the speaker-specific distribution values, the speech signal is authenticated.

A voice query extension method and system. The voice query extension method includes: detecting voice activity of a user from an input signal and extracting a feature vector from the voice activity; converting the feature vector into at least one phoneme sequence and generating the at least one phoneme sequence; matching the at least one phoneme sequence with words registered in a dictionary, extracting a string of the matched words with a linguistic meaning, and selecting the string of the matched words as a query; determining whether the query is in a predetermined first language, and when the query is not in the first language as a result of the determining, converting the query using a phoneme to grapheme rule, and generating a query in the first language; and searching using the query in the first language.

A system and method for rescoring the N-best hypotheses from an automatic speech recognition system by comparing an original speech waveform to synthetic speech waveforms that are generated for each text sequence of the N-best hypotheses. A distance is calculated from the original speech waveform to each of the synthesized waveforms, and the text associated with the synthesized waveform that is determined to be closest to the original waveform is selected as the final hypothesis. The original waveform and each synthesized waveform are aligned to a corresponding text sequence on a phoneme level. The mean of the feature vectors which align to each phoneme is computed for the original waveform as well as for each of the synthesized hypotheses. The distance of a synthesized hypothesis to the original speech signal is then computed as the sum over all phonemes in the hypothesis of the Euclidean distance between the means of the feature vectors of the frames aligning to that phoneme for the original and the synthesized signals. The text of the hypothesis which is closest under the above metric to the original waveform is chosen as the final system output.

A method and apparatus for ranking a plurality of pages identified during a search of a linked database includes forming a linear combination of two or more matrices, and using the coefficients of the eigenvector of the resulting matrix to rank the quality of the pages. The matrices includes information about the pages and are generally normalized, stochastic matrices. The linear combination can include attractor matrices that indicate desirable or “high quality” sites, and / or non-attractor matrices that indicate sites that are undesirable. Attractor matrices and non-attractor matrices can be used alone or in combination with each other in the linear combination. Additional bias toward high quality sites, or away from undesirable sites, can be further introduced with probability weighting matrices for attractor and non-attractor matrices. Other known matrices, such as a co-citation matrix or a bibliographic coupling matrix, can also be used in the present invention.

Documents from a data stream are clustered by first generating a feature vector for each document. A set of cluster centroids (e.g., feature vectors of their corresponding clusters) are retrieved from a memory based on the feature vector of the document using a locality sensitive hashing function. The centroids may be retrieved by retrieving a set of cluster identifiers from a cluster table, the cluster identifiers each indicative of a respective cluster centroid, and retrieving the cluster centroids corresponding to the retrieved cluster identifiers from a memory. Documents may then be clustered into one or more of the candidate clusters using distance measures from the feature vector of the document to the cluster centroids.

Portions from time-domain speech segments are extracted. Feature vectors that represent the portions in a vector space are created. The feature vectors incorporate phase information of the portions. A distance between the feature vectors in the vector space is determined. In one aspect, the feature vectors are created by constructing a matrix W from the portions and decomposing the matrix W. In one aspect, decomposing the matrix W comprises extracting global boundary-centric features from the portions. In one aspect, the portions include at least one pitch period. In another aspect, the portions include centered pitch periods.

A system and method for distinguishing human input events from malware-generated events includes one or more central processing units (CPUs), one or more input devices and memory. The memory includes program code that when executed by the CPU causes the CPU to obtain a first set of input events from a user utilizing the input device. The first input events are used to obtain or derive a feature indicative of the user, such as a multi-dimensional feature vector as provided by a support vector machine. Second input events are then obtained, and the second input events are classified against the feature to determine if either the user or malware initiated the second input events.

The present invention is directed towards systems and methods for trust propagation. The method according to one embodiment comprises calculating a first feature vector for a first user, calculating a second feature for a second user and comparing the first feature vector with the second feature vector to calculate a similarity value. A determination is made as to whether the similarity value falls within a threshold. If the similarity value falls within the threshold, a relationship is recorded between the first user and the second user in a first user profile and a second user profile.

An audio fingerprint is extracted from an audio sample, where the fingerprint contains information that is characteristic of the content in the sample. The fingerprint may be generated by computing an energy spectrum for the audio sample, resampling the energy spectrum logarithmically in the time dimension, transforming the resampled energy spectrum to produce a series of feature vectors, and computing the fingerprint using differential coding of the feature vectors. The generated fingerprint can be compared to a set of reference fingerprints in a database to identify the original audio content.

This invention is a system and method to perform categorization (classification) of multimedia items. These items are comprised of a multitude of disparate information sources, in particular, visual information and textual information. Classifiers are induced based on combining textual and visual feature vectors. Textual features are the traditional ones, such as, word count vectors. Visual features include, but are not limited to, color properties of key intervals and motion properties of key intervals. The visual feature vectors are determined in such a fashion that the vectors are sparse. The vector components are features such as the absence or presence of the color green in spatial regions and the absence or the amount of visual flow in spatial regions of the media items. The text and the visual representation vectors are combined in a systematic and coherent fashion. This vector representation of a media item lends itself to well-established learning techniques. The resulting system, subject of this invention, categorizes (or classifies) media items based both on textual features and visual features.

A method of generating an avatar for a user may include receiving image data of a user from a camera, generating feature vectors for a plurality of features of a user, associating the user with a likely user group selected from a number of defined user groups based on the feature vectors, and assigning an avatar based on the associated user group.

Portions from segment boundary regions of a plurality of speech segments are extracted. Each segment boundary region is based on a corresponding initial unit boundary. Feature vectors that represent the portions in a vector space are created. For each of a plurality of potential unit boundaries within each segment boundary region, an average discontinuity based on distances between the feature vectors is determined. For each segment, the potential unit boundary associated with a minimum average discontinuity is selected as a new unit boundary.

Pattern recognition, wherein a sequence of feature vectors is formed from a digitized incoming signal, the feature vectors comprising feature vector components, and at least one feature vector is compared with templates of candidate patterns by computing a distortion measure. A control signal based on at least one time-dependent variable of the recognition process is formulated, and the distortion measure is computed using only a subset of the vector components of the feature vector, the subset being chosen in accordance with said control signal. This reduces the computational complexity of the computation, as the dimensionality of the vectors involved in the computation is effectively reduced. Although such a dimension reduction decreases the computational need, it has been found not to significantly impair the classification performance.

A method of recognizing a handwritten word of cursive script comprising providing a template of previously classified words, and optically reading a handwritten word so as to form an image representation thereof comprising a bit map of pixels. The external pixel contour of the bit map is extracted and the vertical peak and minima pixel extrema on upper and lower zones respectively of this external contour are detected. Feature vectors of the vertical peak and minima pixel extrema are determined and compared to the template so as to generate a match between the handwritten word and a previously classified word. A method for classifying an image representation of a handwritten word of cursive script is also provided. Also provided is an apparatus for recognizing a handwritten word of cursive script. This apparatus comprises a template of previously classified words, a reader for optically reading a handwritten word, and a controller being linked thereto for generating a match between the handwritten word and the previously classified words. Two algorithms are provided for respectively correcting the skew and slant of a word image.

A system and method of testing and tuning a speech recognition system by providing pronunciations to the speech recognizer. First a text document is provided to the system and converted into a sequence of phonemes representative of the words in the text. The phonemes are then converted to model units, such as Hidden Markov Models. From the models a probability is obtained for each model or state, and feature vectors are determined. The feature vector matching the most probable vector for each state is selected for each model. These ideal feature vectors are provided to the speech recognizer, and processed. The end result is compared with the original text, and modifications to the system can be made based on the output text.

The invention relates to the technology of a knowledge atlas, and aims to provide a Chinese medical knowledge atlas construction method based on deep learning. The Chinese medical knowledge atlas construction method comprises the following steps: obtaining relevant data of a medical field from a data source; using a word segmentation tool to carry out word segmentation on unstructured data, and using an RNN (Recurrent Neural Network) to finish a sequence labeling task to identify entities related to medical care, so as to realize the extraction of knowledge units; carrying out feature vector construction on the entity, and utilizing the RNN to carry out sequence labeling and finish the identification of a relationship among the knowledge units; carrying out entity alignment, and then utilizing the extracted entities and the relationship between the entities to construct the knowledge atlas. According to the Chinese medical knowledge atlas construction method, a recurrent neural network is artfully used for extracting the knowledge units and identifying the relationship among the knowledge units so as to favorably finish the processing of the unstructured data. According to the Chinese medical knowledge atlas construction method, features suitable for the medical care field are put forward to carry out a training task of a network. Compared with general features, the features put forward by the method can better represent a medical entity, and therefore, the relationship among the extracted knowledge units can be more accurate and comprehensive.

A method of and system for transient noise rejection for improved speech recognition. The method comprises the steps of (a) receiving audio including user speech and at least some transient noise associated with the speech, (b) converting the received audio into digital data, (c) segmenting the digital data into acoustic frames, and (d) extracting acoustic feature vectors from the acoustic frames. The method also comprises the steps of (e) evaluating the acoustic frames for transient noise on a frame-by-frame basis, (f) rejecting those acoustic frames having transient noise, (g) accepting as speech frames those acoustic frames having no transient noise and, thereafter, (h) recognizing the user speech using the speech frames.

Feature vectors representing each of a plurality of overlapping frames of an arbitrary, text independent speech signal are computed and compared to vector parameters and variances stored as codewords in one or more codebooks corresponding to each of one or more enrolled users to provide speaker dependent information for speech recognition and / or ambiguity resolution. Other information such as aliases and preferences of each enrolled user may also be enrolled and stored, for example, in a database. Correspondence of the feature vectors may be ranked by closeness of correspondence to a codeword entry and the number of frames corresponding to each codebook are accumulated or counted to identify a potential enrolled speaker. The differences between the parameters of the feature vectors and codewords in the codebooks can be used to identify a new speaker and an enrollment procedure can be initiated. Continuous authorization and access control can be carried out based on any utterance either by verification of the authorization of a speaker of a recognized command or comparison with authorized commands for the recognized speaker. Text independence also permits coherence checks to be carried out for commands to validate the recognition process.

An object of the present invention is to provide a device and a method for recognizing hand shape and position even if a hand image to be provided for recognition is rather complicated in shape, and a recording medium having a program for carrying out the method recorded thereon.A hand image normalization part 11 deletes a wrist region respectively from a plurality of images varied in hand shape and position before subjecting the images to normalization in hand orientation and size to generate hand shape images. An eigenspace calculation part 13 calculates an eigenvalue and an eigenvector respectively from the hand shape images under an analysis based on an eigenspace method. An eigenspace projection part 15 calculates eigenspace projection coordinates by projecting the hand shape images onto an eigenspace having the eigenvectors as a basis. A hand image normalization part 21 deletes a wrist region from an input hand image, and generates an input hand shape image by normalizing the input hand image to be equivalent to the hand shape images. An eigenspace projection part 22 calculates eigenspace projection coordinates for the input hand shape image by projecting the same onto the eigenspace having the eigenvectors as the basis. A hand shape image selection part 23 compares the eigenspace projection coordinates calculated for the input hand shape image with each of the eigenspace projection coordinates calculated for the hand shape images, and then determines which of the hand shape images is closest to the input hand shape image. A shape / position output part 24 outputs shape information and position information on the determined hand shape image.

Methods of imaging objects based on wavelet retrieval of scenes utilize wavelet transformation of plural defined regions of a query image. By increasing the granularity of the query image to greater than one region, accurate feature vectors are obtained that allow for robust extraction of corresponding regions from a database of target images. The methods further include the use of sliding windows to decompose the query and target images into regions, and the clustering of the regions utilizing a novel similarity metric that ensures robust image matching in low response times.

A method and computer-readable medium are provided for identifying clean signal feature vectors from noisy signal feature vectors. The method is based on variational inference techniques. One aspect of the invention includes using an iterative approach to identify the clean signal feature vector. Another aspect of the invention includes using the variance of a set of noise feature vectors and / or channel distortion feature vectors when identifying the clean signal feature vectors. Further aspects of the invention use mixtures of distributions of noise feature vectors and / or channel distortion feature vectors when identifying the clean signal feature vectors. Additional aspects of the invention include using a variance for the noisy signal feature vector conditioned on fixed values of noise, channel transfer function, and clean speech, when identifying the clean signal feature vector.

A device capable of achieving recognition at a high accuracy and with fewer calculations and which utilizes an HMM. The present device has a vector quantizing circuit generating a model by quantizing vectors of a training pattern having a vector series, and converting the vectors into a label series of clusters to which they belong, a continuous distribution probability density HMM generating circuit for generating a continuous distribution probability density HMM from a quantized vector series corresponding to each label of the label series, and a label incidence calculating circuit for calculating the incidence of the labels in each state from the training vectors classified in the same clusters and the continuous distribution probability density HMM.

A content-addressable and searchable storage system for managing and exploring massive amounts of feature-rich data such as images, audio or scientific data, is shown. The system comprises a segmentation and feature extraction unit for segmenting data corresponding to an object into a plurality of data segments and generating a feature vector for each data segment; a sketch construction component for converting a feature vector into a compact bit-vector corresponding to the object; a similarity index comprising a plurality of compact bit-vectors corresponding to a plurality of objects; and an index insertion component for inserting a compact bit-vector corresponding to an object into the similarity index. The system may further comprise an indexing unit for identifying a candidate set of objects from said similarity index based upon a compact bit-vector corresponding to a query object. Still further, the system may additionally comprise a similarity ranking component for ranking objects in said candidate set by estimating their distances to the query object.

The invention puts forward an image intelligent mode identification search method. The method can establish an image sample training set database and combine with basic text search engine technology and basic image content inquiry technology, so that a network creeper can perform Internet image search and URL information resolution, so as to catch the image URL and relevant information into a local primary database; perform such pre-processes as preliminary filtration, decompression and image pre-classification and etc for the images; then, calculate color characteristics, grain characteristics and shape characteristics of the extraction images, so as to gain corresponding characteristic vector sets; combine with the image URL information before saving the images into the image basic database and establishing an index for the images; perform characteristic vector similarity calculation for images in the image basic databases and sample training sets, and then, save the classified images into an image classification database; accept key words or image description that are input by the user, create the index vector, perform similarity calculation with the image characteristic vectors in the image classification database, and then, return the index results to the user.

Audio / Visual data is classified into semantic classes such as News, Sports, Music video or the like by providing class models for each class and comparing input audio visual data to the models. The class models are generated by extracting feature vectors from training samples, and then subjecting the feature vectors to kernel discriminant analysis or principal component analysis to give discriminatory basis vectors. These vectors are then used to obtain further feature vector of much lower dimension than the original feature vectors, which may then be used directly as a class model, or used to train a Gaussian Mixture Model or the like. During classification of unknown input data, the same feature extraction and analysis steps are performed to obtain the low-dimensional feature vectors, which are then fed into the previously created class models to identify the data genre.