35481 results about "Training set" patented technology

The invention relates to a method and apparatus for training a multilingual speech model set. The multilingual speech model set generated is suitable for use by a speech recognition system for recognizing spoken utterances for at least two different languages. The invention allows using a single speech recognition unit with a single speech model set to perform speech recognition on utterances from two or more languages. The method and apparatus make use of a group of a group of acoustic sub-word units comprised of a first subgroup of acoustic sub-word units associated to a first language and a second subgroup of acoustic sub-word units associated to a second language where the first subgroup and the second subgroup share at least one common acoustic sub-word unit. The method and apparatus also make use of a plurality of letter to acoustic sub-word unit rules sets, each letter to acoustic sub-word unit rules set being associated to a different language. A set of untrained speech models is trained on the basis of a training set comprising speech tokens and their associated labels in combination with the group of acoustic sub-word units and the plurality of letter to acoustic sub-word unit rules sets. The invention also provides a computer readable storage medium comprising a program element for implementing the method for training a multilingual speech model set.

A text-to-pronunciation system (11) includes a large training set of word pronunciations (19) and an extractor for extracting language specific information from the training set to produce pronunciations for words not in its training set. A learner (13) forms pronunciation guesses for words in the training set and for finding a transformation rule that improves the guesses. A rule applier (15) applies the transformation rule found to guesses. The learner (13) repeats the finding of another rule and the rule applier (15) applies the new rule to find the rules that improves the guesses the most.

A method for segmenting digitized images includes providing a training set comprising a plurality of digitized whole-body images, providing labels on anatomical landmarks in each image of said training set, aligning each said training set image, generating positive and negative training examples for each landmark by cropping the aligned training volumes into one or more cropping windows of different spatial scales, and using said positive and negative examples to train a detector for each landmark at one or more spatial scales ranging from a coarse resolution to a fine resolution, wherein the spatial relationship between a cropping windows of a coarse resolution detector and a fine resolution detector is recorded.

A test apparatus having a base which performs a plurality of tests for different analytes in body fluids. A portable tester, such as a glucose meter, is detachably mounted to the base and provides a convenient and ergonomic hand-held instrument which a diabetic may carry to frequently monitor blood glucose levels. When mounted or docked to the base, the portable tester is data linked to the base. The invention provides a communications network among the base, the portable tester and another device such as a PC. The portable tester can therefore be kept small and inexpensive since advanced computing and data storage capabilities are provided in the base or a PC.

The invention provides a substantially moisture-proof, air-tight apparatus for both dispensing a plurality of diagnostic test strips and testing a biological fluid dispensed onto the strip. One strip may be advanced for use in testing using a single, translational movement.

The present invention provides an apparatus and method for high resolution spectroscopy (approximately 10 picometer wavelength resolution) using a tunable optical filter (TOF) for analyzing a formation fluid sample downhole and at the surface to determine formation fluid parameters. The analysis comprises determination of gas oil ratio, API gravity and various other fluid parameters which can be estimated after developing correlations to a training set of samples using a neural network or a chemometric equation.

The invention discloses a human face super-resolution reconstruction method based on a generative adversarial network and sub-pixel convolution, and the method comprises the steps: A, carrying out the preprocessing through a normally used public human face data set, and making a low-resolution human face image and a corresponding high-resolution human face image training set; B, constructing the generative adversarial network for training, adding a sub-pixel convolution to the generative adversarial network to achieve the generation of a super-resolution image and introduce a weighted type loss function comprising feature loss; C, sequentially inputting a training set obtained at step A into a generative adversarial network model for modeling training, adjusting the parameters, and achieving the convergence; D, carrying out the preprocessing of a to-be-processed low-resolution human face image, inputting the image into the generative adversarial network model, and obtaining a high-resolution image after super-resolution reconstruction. The method can achieve the generation of a corresponding high-resolution image which is clearer in human face contour, is more specific in detail and is invariable in features. The method improves the human face recognition accuracy, and is better in human face super-resolution reconstruction effect.

The present invention relates to a system (10, 200) and methodology (74) to enable a plurality of information associated with electronic messages, for example, to be automatically prioritized by a priorities system (12, 230) for transmittal to a user or system. The priorities system (12,230) can employ classifiers (20) that can be explicitly and/or implicitly trained to prioritize one or more received messages (14) according to a learned importance to the user. As an example, messages (14) can be classified as high, medium, low or other degrees of importance via a training set of examples (30) or types of messages having similar degrees of importance. A background monitor (34) can be provided to monitor a user's activities regarding message processing to further refine or tune the classifier (20) according to the user's personal decisions relating to message importance. Other priorities classifications can involve determinations relating to a loss associated with a time for delayed review or processing of the message.

A system is described for using with fine pitch devices including singulated bare die, semiconductor wafers, chip sized packages, printed circuit boards, and the like to determine that the fine pitch device is not faulty. The system is also usable for transfer of data, energy, for collecting data measurements or measurement-related data between two pieces, and for effecting at least part of an identification process. The disclosed embodiment includes a substrate having a circuit pad pattern in the mirror image of the pattern of contact points, usually bond pads of the fine pitch device to be connected. A conductive elastomeric probe is permanently formed on the circuit pads of the substrate such that the probe is malleable and allows repetitive electrical contact. The system may also contain an alignment template for orienting the fine pitch device onto the elastomeric probes of the contact point pattern of the substrate.

The invention discloses a fundus image retinal vessel segmentation method and a fundus image retinal vessel segmentation system based on deep learning. The fundus image retinal vessel segmentation method comprises the steps of performing data amplification on a training set, enhancing an image, training a convolutional neural network by using the training set, segmenting the image by using a convolutional neural network segmentation model to obtain a segmentation result, training a random forest classifier by using features of the convolutional neural network, extracting a last layer of convolutional layer output from the convolutional neural network, using the convolutional layer output as input of the random forest classifier for pixel classification to obtain another segmentation result, and fusing the two segmentation results to obtain a final segmentation image. Compared with the traditional vessel segmentation method, the fundus image retinal vessel segmentation method uses the deep convolutional neural network for feature extraction, the extracted features are more sufficient, and the segmentation precision and efficiency are higher.

An analyte detection system is provided with calibration information uniquely specific to the set of test strips to which the sample is to be applied. The calibration information may be stored in permanent memory of the testing device, such that the device is discarded after use of all the test strips in a kit, or it may be stored in a calibration chip accompanying the set of test strips and distributed therewith, thereby enabling re-use of the testing device with a different set of test strips and associated calibration chip.

A system for, and method of, text-to-phoneme (TTP) mapping and a digital signal processor (DSP) incorporating the system or the method. In one embodiment, the system includes: (1) a letter-to-phoneme (LTP) mapping generator configured to generate an LTP mapping by iteratively aligning a full training set with a set of correctly aligned entries based on statistics of phonemes and letters from the set of correctly aligned entries and redefining the full training set as a union of the set of correctly aligned entries and a set of incorrectly aligned entries created during the aligning and (2) a model trainer configured to update prior probabilities of LTP mappings generated by the LTP generator and evaluate whether the LTP mappings are suitable for training a decision-tree-based pronunciation model (DTPM).

A method and a system for making merging decisions for a translation are disclosed which are suited to use where the target language is a productive compounding one. The method includes outputting decisions on merging of pairs of words in a translated text string with a merging system. The merging system can include a set of stored heuristics and/or a merging model. In the case of heuristics, these can include a heuristic by which two consecutive words in the string are considered for merging if the first word of the two consecutive words is recognized as a compound modifier and their observed frequency f₁ as a closed compound word is larger than an observed frequency f₂ of the two consecutive words as a bigram. In the case of a merging model, it can be one that is trained on features associated with pairs of consecutive tokens of text strings in a training set and predetermined merging decisions for the pairs. A translation in the target language is output, based on the merging decisions for the translated text string.

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.

A method and apparatus of diagnosing a cardiac disease state in as little as two minutes involving the utilization of an evanescent wave assay system in conjunction with a data acquisition and analysis procedure that monitors the precision of assay results in real time (i.e., while data is being acquired). The method includes diagnosing a disease state using a diagnostic procedure (e.g., an immunoassay) wherein the testing device informs the person conducting the test of the results of the test as soon as reliable test data is obtained (generally, <5% variation in the reaction rate of the assay). After which point, the diagnostic procedure may be terminated.

The invention discloses a convolution neural network parallel processing method based on a large-scale high-performance cluster. The method comprises the steps that (1) a plurality of copies are constructed for a network model to be trained, model parameters of all the copies are identical, the number of the copies is identical with the number of nodes of the high-performance cluster, each node is provided with one model copy, one node is selected to serve as a main node, and the main node is responsible for broadcasting and collecting the model parameters; (2) a training set is divided into a plurality of subsets, the training subsets are issued to the rest of sub nodes except the main mode each time to conduct parameter gradient calculation together, gradient values are accumulated, the accumulated value is used for updating the model parameters of the main node, and the updated model parameters are broadcast to all the sub nodes until model training is ended. The convolution neural network parallel processing method has the advantages of being capable of achieving parallelization, improving the efficiency of model training, shortening the training time and the like.

The present invention is a method, circuit and system for determining a reference voltage. Some embodiments of the present invention relate to a system, method and circuit for establishing a set of operating reference cells to be used in operating (e.g. reading) cells in an NVM block or array. As part of the present invention, at least a subset of cells of the NVM block or array may be read using each of two or more sets of test reference cells, where each set of test reference cells may generate or otherwise provide reference voltages at least slightly offset from each other set of test reference cells. For each set of test reference cells used to read at least a subset of the NVM block, a read error rate may be calculated or otherwise determined. A set of test reference cells associated with a relatively low read error rate may be selected as the set of operating reference cells to be used in operating (e.g. reading) other cells, outside the subset of cells, in the NVM block or array. In a further embodiment, the selected set of test reference cells may be used to establish an operating set of reference cells having reference voltages substantially equal to those of the selected test set.

The invention solves the inconvenience that a customer would experience when his/her electronic wallet was lost or broken down. A backup condition testing means tests predefined backup conditions to determine whether to make a backup, when an electronic wallet device communicates with any other device to transfer monetary data. When the backup condition testing means has determined to make a backup, a backup data sending means sends a predetermined set of backup data collected from the electronic wallet device. A backup data management means receives the backup data from the backup data sending means and saves the received data into its backup database. In this way, the proposed system automatically saves backup data of the electronic wallet device when any backup condition is met. This mechanism permits the customer to have an alternative electronic wallet device to be promptly issued when his/her electronic wallet device is lost.

A fault detection apparatus according to an embodiment is provided with a measurement unit, a memory unit, a control unit, and a display unit. The measurement unit measures a first time period taken until a reflection signal reflected on a fault of a test apparatus is received after a first signal is transmitted to the test apparatus. The memory unit includes a CAD data unit having CAD data of the test apparatus, and a model data unit to store model data indicating a relation between the first time period and a predicted conduction distance of the first signal according to the CAD data. The control unit calculates a range of a test object selected in the test apparatus based on the CAD data, calculates the predicted conduction distance from the first time period based on the model data, and specifies a position of a fault of the test apparatus which is separated by the predicted conduction distance from the measurement unit in the range of the test object. The display unit displays the position of the fault in the CAD data.

Heart valve leaflet selection methods and apparatuses which subject individual leaflets to loads and measure the resulting deflection to more reliably group leaflets of similar physical characteristics for later assembly in prosthetic heart valves. The deflection testing may be accomplished using a variety of test set ups which are designed to impart a load on the leaflet which simulates the actual loading within a heart valve. The results from a number of deflection tests are used to categorize individual leaflets, which data can be combined with other data regarding the characteristics of the leaflet to better select leaflets for assembly into a multi-leaflet heart valve. In one embodiment, the deflection test is combined with an intrinsic load test, and leaflets having similar deflection and intrinsic load values used in the same heart valve. One apparatus for testing the leaflets includes a frame for securing the arcuate cusp of the leaflet while the straight coapting edge remains free, to simulate the actual leaflet mounting configuration within the heart valve prosthesis. The frame may include a lower portion having a recess for the leaflet and plurality of receptor holes around the peripheral edge of the recess, and an upper portion having a plurality of needles which extend downward through the leaflet and into the receptor holes and secure the edges of the leaflet.

An apparatus and method provide non-intrusive in-service testing of audio/video synchronization testing without using traditional audio marker tones. The network includes an A/V synchronous test signal generator which injects video and audio markers into the video and audio non-intrusively and routes the two signals into a switch where they are switched into a channel for encoding and transmission via the ATM network. At the distant end the signal is decoded and routed by a switch into the A/V test generator and measurement set where the markers are detected and the A/V skew calculated, after which the audio and video are routed to the subscriber. The A/V test set signal generator includes a Video Blanking Interval (VBI) test signal generator and a white noise generator, the former injecting a marker into the video signal and the later injecting an audio marker into the audio signal. The video marker is injected into the VBI and broadband, background audio noise to measure the delay between the audio and video components of a broadcast. The marking of the audio is accomplished by gradually injecting white noise into the audio channel until the noise level is 6 dB above the noise floor of the audio receiver. As a precursor A/V sync signal, a small spectrum of the white noise is notched or removed. This signature precludes inadvertent recognition of program audio noise as the audio marker.