Abstract: A system and method of image classification may include linking between a plurality of predetermined objects and categories, assigning a set of attributes for each object until relationships between all categories are determined, storing at least one classified image for each object in an object database, receiving an unclassified image, comparing the received unclassified image to the at least one classified image, determining at least one attribute of the received unclassified image, based on the comparison and based on the assigned attributes for each object, assigning a classification of the received unclassified image, with correlation between the at least one determined attribute of the received unclassified image, and relationships between the categories.

Abstract: Disclosed are an apparatus and a method for detection of foreign substances in a depth sensing system. In one embodiment, a depth sensing device includes a light source to emit light, an image sensor and a processor. The image sensor receives through an optical component the light reflected by environment of the depth sensing device. The image sensor further generates a depth map including a plurality of pixel values corresponding to distances between the depth sensing device and the environment. The processor detects a blurred portion of the depth map due to a presence of a foreign substance on the optical component. The processor may further cause outputting a user alert of the presence of the foreign substance on the optical component.

Abstract: Systems and methods are disclosed herein for automatically identifying a query object within a visual medium. The technique generally involves receiving as input to a neural network a query object and a visual medium including the query object. The technique also involves generating, by the neural network, representations of the query object and the visual medium defining features of the query object and the visual medium. The technique also involves generating, by the neural network, a heat map using the representations. The heat map identifies a location of pixels corresponding to the query object within the visual medium and is usable to generate an updated visual medium highlighting the query object.

Abstract: Methods, apparatuses and systems for classifying defects for a defect inspection system are disclosed. The defect inspection system can be used to inspect and manage wafer or reticle defects. The method includes receiving a defect record based on an inspection of a target specimen, the defect record comprising a defect image associated with an unknown defect, selecting, by a computing device using a first processing unit, components ranked by significance from the defect image using a first learning technique, and determining, by the computing device using the first processing unit, whether the defect image is associated with a known defect type based on the components ranked by significance using a second learning technique.

Abstract: Disclosed herein are systems, devices, and methods for detecting anomalies in multivariate data received from an asset-related data source, such as signal data and/or other data from an asset. According to an example, a platform may receive multivariate data from an asset in an original coordinate space and transform the data in the original coordinate space to a transformed coordinate space having a relatively fewer number of dimensions. Additionally, the platform may standardize the data in the transformed coordinate space and modify the standardized data based on a comparison between the standardized data and a set of threshold values previously defined via training data reflective of normal asset operation. Thereafter, the platform may inversely transform the modified data back to the original coordinate space and perform an analysis to detect anomalies.

Abstract: Embodiments of the present disclosure disclose an image character recognition model generation method and apparatus, and a vertically-oriented character image recognition method and apparatus. The image character recognition model generation method includes: generating a rotated line character training sample, wherein the rotated line character training sample includes a rotated line character image and an expected character recognition result corresponding to the rotated line character image, and there is a difference of 90 degrees between character units in the rotated line character image and character units in a standard line character image; and training a set neural network by using the rotated line character training sample, to generate an image character recognition model.

Abstract: Provided are techniques for image text analysis for identifying hidden text. An Optical Character Reader (OCR) is utilized to extract a text string from an image. Context within the image is analyzed. It is determined that the extracted text string is a partial text string based on the context. For a first radius level of a plurality of radius levels, a segmented sub-image is identified around the partial text string within the first radius level, an image search on the segmented sub-image is performed to identify a candidate text string, and, in response to determining that the candidate text string is a complete text string, the complete text string is provided for performing an action.

Abstract: A system comprises a processor configured to execute instructions to receive an indication of an occurrence of a human gesture and to perform an analysis of the indication of the occurrence of the human gesture to determine contextual criteria having a relationship to the occurrence of the human gesture. The processor may determine a meaning of the human gesture based at least in part on the contextual criteria and a plurality of possible intended meanings for the human gesture. The processor also may execute an instruction respective to determining the meaning of the human gesture, wherein at least a portion of the instruction is dependent upon the meaning of the human gesture.

Abstract: A sheet conveyance apparatus includes an inner main unit, a guiding portion, a conveyance roller pair, a mounting portion, and an urging member. A pair of bearing portions supports a shaft portion of a driven roller. A supporting portion is located on an upstream side or a downstream side in a conveyance direction in the pair of bearing portions and supports a pair of extending portions from below. A pressing portion abuts on the pair of extending portions, from above, between the pair of bearing portions and the supporting portion to elastically deform the pair of extending portions. A regulating portion is located between the pair of bearing portions and the supporting portion in the conveyance direction and regulates an interval between the pair of extending portions such that the interval becomes wider than a length of a roller portion in a width direction.

Abstract: A method of image processing includes: reading a data file in a predetermined page description language including a search key; chaining a plurality of pieces of resource data in each of a plurality of resource dictionaries; loading the chained plurality of pieces of resource data in each of the plurality of resource dictionaries in a memory area including unique keys; searching the requested resource data from the chained pieces of resource data using the search key for matching the unique key of the requested resource data one by one along the chain in both direction from a starting piece of resource data; labeling each of the plurality of pieces of resource data based on a history of the matching. The searching includes determining the starting piece of resource data in the chained plurality of pieces of resource data based on the history.

Abstract: A method of generating transformed print jobs includes inputting a page description language print job and receiving a selection of one or more transformation features and one or more transformation feature settings for each selected transformation feature. For each transformation feature selected, all dependent transformation features are identified. For each dependent transformation feature identified, all dependent feature settings are identified. All combinations of transformation features, transformation feature settings, dependent transformation features, and dependent feature settings are determined. For each combination, the inputted print job is transformed according to the transformation features, transformation feature settings, dependent transformation features, and dependent feature settings of the combination and a transformed print job is outputted.

Abstract: An image processing unit includes a region separation device and a spatial filtering device. The region separation device detects line widths of a manuscript image in which the line widths are different between a vertical line and a horizontal line and determines which of the vertical line or the horizontal line is thinner. The spatial filtering device executes an emphasizing process to make a widening ratio of a thinner line of the vertical line or the horizontal line higher than a widening ratio of the other line.

Abstract: Controlling execution of image processing that adjusts an amount of a printing material applied to the printing medium based on an amount of change of pixel values of image data.

Abstract: According to another embodiment, a system includes a driver circuit that drives a first output and a second output; a coil coupled between the first output and the second output such that the driver circuit drives current through the coil in response to control signals; and a programmable slew circuit coupled to the driver circuit. In some embodiments, a switch is coupled between the first output and the coil. In some embodiments an over-voltage protection circuit is coupled to protect the driver circuit.

Abstract: A RFID transponder includes an active load modulation unit and an energy harvesting unit coupled to the active load modulation unit. The active load modulation unit performs active load modulation on transmitted signals. The energy harvesting unit harvests RF energy from the ambient environment, converts the RF energy to DC energy, stores the DC energy, and supplies the DC energy to the active load modulation unit.

Abstract: A transponder with a disrupting signal source to shut down the transponder. The transponder may use an RFID chip. The disrupting signal source can be an electrical signal source electrically coupled to the RFID chip, for example capacitively coupled. The disruptive signal source can also be a radiation source, for example an infrared source. The disrupting signal source can be connected to a receiver to allow the transponder to be shut down remotely. Multiple transponders as described can be combined to form a multi-transponder device in which transponders can be shut down selectively.

Abstract: An electronic badge comprises a housing having a form factor of a mobile device, a processor, and memory coupled to the processor and including an observation and feedback application. Further, the electronic badge comprises a first transceiver that communicates on a first personal-area network, a second transceiver that communicates on a second personal-area network, and a third transceiver that communicates on a local area network. An information feedback device is controlled by the processor to activate to provide information to the user that feedback has been either given or received, wherein the information feedback device is operated by the processor in response to the processor executing the observation and feedback application, and in response to the processor interacting with at least one of the first transceiver, the second transceiver, and the third transceiver.

Abstract: The application discloses a single-chip financial audio IC card, including a security chip, a contact module, a non-contact module and an audio module switch. The security chip is used for controlling, analyzing and processing data, which is shared by a controlling, analyzing and processing data module; the contact module has function of contact communication, including a 7816 interface; the non-contact module has function of non-contact communication, including a RF coil; the audio module has function of acoustic communication, including a switch, a battery and a buzzer; the contact module, non-contact module and audio module are connected in parallel to the security chip. The application of single-chip financial audio IC card has the advantages of low cost, small processing difficulty, good compatibility, high security, good stability and reliability.

Abstract: In a wireless communication device, radiation conductors including a first and second end portions are reformed on an upper surface of a radiation conductor base material. First and second terminal electrodes are provided at a same or substantially the same interval as the first and second end portions, on a lower surface of a RFIC element. A seal includes an adhesive surface larger than a principal surface of the RFIC element. The RFIC element is arranged on the upper surface of the radiation conductor substrate so that each of the first and second terminal electrodes comes into contact with the first and second end portions. The seal is pasted to the radiation conductor substrate so as to cover the RFIC element.

Abstract: An RF tag that includes an inlay (made of an IC chip and an antenna) and a housing that houses the inlay therein. Furthermore, the housing comprises a top cover that covers at least an upper surface side of the inlay, the top cover is made of a predetermined metal material, the top cover includes a cutaway that opens in the upper surface side of the inlay, and the top cover is electrically connected to the inlay to function as an auxiliary antenna of the inlay.

Abstract: Embodiments may provide methods by which fusion of information may be applied to uncertainty reduction in the results of classifier-based data analysis. For example, a method for data analysis may comprise generating a plurality of sets of data samples, each set of data samples representing a portion of input data at plurality of scales, each data sample in a set may represent the portion of the input data at a different scale and location, generating a feature map from each data sample of at least one set of data samples by learning and aggregating features using a first multi-layer convolutional processing, each data sample may be processed with multi-layer convolutional processing separately from other data samples, and generating a feature map by combining the feature maps from the data samples of each set of data samples by performing multiple-scale-multiple-location label fusion using a second multi-layer convolutional processing.

Abstract: An apparatus and a method. The apparatus includes a first recurrent network in a first layer; a second recurrent network in a second layer connected to the first recurrent network; a distant input gate connected to the second recurrent network; a first highway gate connected to the distant input gate and the second recurrent network; a first elementwise product projection gate connected to the distant input gate, the highway gate, and the second recurrent network; a second highway gate connected to the first recurrent network and the second recurrent network; and a second elementwise product projection gate connected to the first recurrent network, the second highway gate, and the second recurrent network.

Abstract: An apparatus and a method. The apparatus includes a plurality of long short term memory (LSTM) networks, wherein each of the plurality of LSTM networks is at a different network layer, wherein each of the plurality of LSTM networks is configured to determine a residual function, wherein each of the plurality of LSTM networks includes an output gate to control what is provided to a subsequent LSTM network, and wherein each of the plurality of LSTM networks includes at least one highway connection to compensate for the residual function of a previous LSTM network.

Abstract: Provided are a machine learning method based on weakly supervised learning includes extracting feature maps about a dataset given a first type of information and not given a second type of information by using a convolutional neural network (CNN), updating the CNN by back-propagating a first error value calculated as a result of performing a task corresponding to the first type of information by using a first model, and updating the CNN by back-propagating a second error value calculated as a result of performing the task corresponding to the first type of information by using a second model different from the first model, wherein the second type of information is extracted when the task corresponding to the first type of information is performed using the second model.

Abstract: A method includes: training a first neural network using a first training dataset; inputting each test data of a first test dataset to the first neural network; calculating output data of the first neural network for each test data of the first test dataset; composing a second training dataset of training data from the first test dataset that causes the first neural network to output data within a first range; and training a second neural network using the second training dataset.

Abstract: This disclosure describes techniques and architectures to morph well-trained networks to other related applications or modified networks with relatively little retraining. For example, a well-trained neural network (e.g., parent network) may be morphed to a new neural network (e.g., child network) so that the new neural network function may be preserved. After morphing a parent network, the child network may inherit the knowledge from its parent network and also may have a potential to continue growing into a more powerful network. Such morphing and growing may occur with a relatively short training time.

Abstract: A machine learning module may generate a probability distribution from training data including labeled modeling data correlated with reflection data. Modeling data may include data from a LIDAR system, camera, and/or a GPS for a target environment/object. Reflection data may be collected from the same environment/object by a radar and/or an ultrasonic system. The probability distribution may assign reflection coefficients for radar and/or ultrasonic systems conditioned on values for modeling data. A mapping module may create a reflection model to overlay a virtual environment assembled from a second set of modeling data by applying the second set to the probability distribution to assign reflection values to surfaces within the virtual environment. Additionally, a test bench may evaluate an algorithm, for processing reflection data to generate control signals to an autonomous vehicle, with simulated reflection data from a virtual sensor engaging reflection values assigned within the virtual environment.

Abstract: A device, system, product and method of controlling resistive processing units (RPUs), includes applying an input voltage signal to each node of an array of resistive processing units, and controlling a learning rate of the array of resistive processing units by varying an amplitude of the input voltage signal to the array of resistive processing units. A conductance state of the array of resistive processing units is varied according to the amplitude received at each of the resistive processing units of the array of resistive processing units. The controlling of the amplitude of input voltage signal is according to a processor of a control device.

Abstract: An encoder and decoder for translating sequential data into a fixed dimensional vector are created by applying an encoding-trainer input vector set as input to an encoding neural network to generate an encoding-trainer output vector set. One vector of the encoding-trainer output vector set is selected and a decoding-trainer input vector set is generated from it. A decoding neural network is trained by applying the generated decoding-trainer input vector set to the decoding neural network. The encoder and decoder can be used in implementations processing sequential data.

Abstract: A deep embedding forest-based (DEF) model for improving on-line serving time for classification learning methods and other tasks such as, for example, predicting user selection of search results provided in response to a query or for image, speech or text recognition. Initially, a deep neural network (DNN) model is trained to determine parameters of an embedding layer, a stacking layer, deep layers and a scoring layer thereby reducing high dimensional features. After training the DNN model, the parameters of the deep layers and the scoring layer of the DNN model and discarded and the parameters of the embedding layer and the stacking layer are extracted. The extracted parameters from the DNN model then initialize parameters of an embedding layer and a stacking layer of the DEF model such that only a forest layer of the DEF model is then required to be trained. Output from the DEF model is stored in computer memory.

Abstract: A learning apparatus and method for learning a model corresponding to time-series input data, comprising: acquire the time-series input data; supply a plurality of input nodes of the model with a plurality of input values corresponding to input data at one time point in the time-series input data; store values of hidden nodes; compute a conditional probability of each input value at the one time point on a condition that an input data sequence has occurred, based on the input data sequence before the one time point in the time-series input data, on the stored values of hidden nodes, and on weight parameters; and perform a learning process that further increases a conditional probability of input data occurring at the one time point on the condition that the input data sequence has occurred, by adjusting the weight parameters.

Abstract: Methods and systems for language processing includes initializing a word embedding matrix based on pre-determined word classes, such that matrix entries associated with a class of which a word is a member are initialized to a non-zero value and other entries are initialized to zero. A neural network is trained based on the initialized word embedding matrix to generate a neural network language model. A language processing task is performed using the neural network language model.

Abstract: A computer-implemented method is provided for neural network training. The method includes improving a cache utilization by one or more processors during multiple training stages of a neural network, by performing a stage-wise mini-batch process on a set of samples used for the multiple training stages. The stage-wise mini-batch process waits for each of the multiple training stages to complete using a system wait primitive to improve the cache utilization.

Abstract: A method for personalized intelligent wake-up system based on multimodal deep neural network comprises monitoring a sleeping status of a user; obtaining a current sleeping-stage of the user within a current time frame and a prediction of a next sleeping-stage of the user for a next time frame; correcting the current sleeping-stage of the user through combining the current sleeping-stage and the prediction of the next sleeping-stage; determining a wake up strategy for the current time frame; determining a relationship between each of a plurality of alarm impulses adopted to wake up the user and a corresponding reaction of the user; identifying a change in the current sleeping-stage for the current time frame; determining an alarm impulse to be triggered for waking up the user; and triggering the determined alarm impulse.

Abstract: A technique for assigning confidence scores to relationship entries in a knowledge graph includes assigning respective initial confidence scores to relationship n-tuples in a knowledge graph. Each of the relationship n-tuples designates at least a first entity, a second entity, and a relationship between the first and second entities or a single entity and a relationship between the single entity and one or more properties of the single entity. Respective feature vectors are associated with each of the relationship n-tuples. A training set that includes at least a subset of the relationship n-tuples labeled with respective ground truth labels is generated. Respective initial weights are learned for the feature vectors based on the training set. Respective subsequent confidence scores are generated for each of the relationship n-tuples based on the initial weights for the feature vectors.

Abstract: A technique for responding to user input includes assigning respective initial confidence scores to relationship n-tuples in a knowledge graph (KG). Each of the n-tuples designates at least a first entity, a second entity, and a relationship between the first and second entities or a single entity and a relationship between the single entity and one or more properties of the single entity. Respective feature vectors are associated with each of the n-tuples. A training set that includes at least a subset of the n-tuples labeled with respective ground truth labels is generated. Respective initial weights are learned for the feature vectors based on the training set. Respective subsequent confidence scores are generated for each of the n-tuples based on the initial weights for the feature vectors. A response to user input is generated based on the subsequent confidence scores for one or more of the n-tuples.

Abstract: Embodiments of the present invention provide systems and methods for predicting personality attributes. The method includes accessing a set of data for an operator. The method further includes analyzing the set of data, matching the operator to a set of personality attributes, and storing the matched set of personality attributes for the operator.

Abstract: Systems, methods, and non-transitory computer-readable media can determine a respective latent representation for each entity in a set of entities that are accessible through the social networking system, wherein a latent representation for an entity is determined based at least in part on a topic model associated with the entity, each latent representation for an entity having a lower dimensionality than a topic model of the entity. One or more candidate entities that are related to a first entity can be determined based at least in part on the respective latent representations for the candidate entities and the first entity. At least a first candidate entity from the one or more candidate entities can be provided as a recommendation to a user that formed a connection with the first entity.

Abstract: Methods, computer program products, and systems are presented. The methods include, for instance: building model connections between models in a knowledgebase, which stores case data as model networks. An exploration probability stored in the knowledgebase indicates a likelihood of new connections based on a case data input to be employed for an analytical model of the case data input, which includes numerous stages and multiple model choices in each stage. Based on the new connections and model networks of the knowledgebase, paths are created and performance of respective paths/connections are evaluated. A predefined number of top performing paths are selected and the models and attributes that do not appear in the top performing paths are eliminated from the knowledgebase. Probabilities of model networks and a future case data input are updated accordingly and a best-fit model for the case data input is presented to a user.

Abstract: A method for determining the predictive value of a feature may include: (a) performing predictive modeling procedures associated with respective predictive models, wherein performing each modeling procedure includes fitting the associated model to an initial dataset representing an initial prediction problem; (b) determining a first accuracy score of each of the fitted models, representing an accuracy with which the fitted model predicts an outcome of the initial prediction problem; (c) shuffling values of a feature across observations included in the initial dataset, thereby generating a modified dataset representing a modified prediction problem; (d) determining a second accuracy score of each of the fitted models, representing an accuracy with which the fitted model predicts an outcome of the modified prediction problem; and (e) determining a model-specific predictive value of the feature for each of the fitted models based on the first and second accuracy scores of the fitted model.

Abstract: A system, a machine-readable storage medium storing instructions, and a computer-implemented method as described herein are directed to a Fast Ranker Engine that identifies global model features present in an article in a social network service. The Fast Ranker Engine assembles respective fixed vectors based on at least one member account feature and each coefficient that corresponds to a present global article feature of the global model. The Fast Ranker Engine generates a transformation feature(s) for a prediction model of the article based on the respective fixed vectors.

Abstract: T-shirt size recommendation for a first user based on crowd sourcing. A recommendation for t-shirt size may be made by defining a first image of a container based on historical data related to container instantiation; receiving, by the computer, from a first user, a request to deploy a first instantiation of a container corresponding to the first image, with the request including input data including information indicative of a planned context of the first instantiation; applying a plurality of machine logic based rules to the input data to determine a first a first recommended T-shirt size associated with the first image; and communicating, by the computer, the first recommended T-shirt size to the first user for use with the first instantiation of the container corresponding to the first image.

Abstract: A data conversion method performed by a computer, the method is configured to cause the computer to; when the computer receives a plurality of data, acquire first information and second information from a storage, the first information designating any data among the plurality of data, the second information designating a predetermined number of data, data formats of which are collectively converted among the plurality of data; acquire, based on the acquired first information and the acquired second information, the predetermined number of conversion target data from the received plurality of data; extract each pieces of specific information embedded in each of the acquired predetermined number of conversion target data; and convert the data format of each of the respective predetermined number of conversion target data by using the extracted predetermined number of pieces of the specific information.

Abstract: Methods and systems are provided for cloud processing data streamed from a vehicle and a home (e.g., any location) associated with a user account. One method includes receiving a data stream from the vehicle entity, where the data stream from the vehicle entity includes metadata from one or more data producing objects of the vehicle entity. And, receiving a data stream from the home entity, where the data stream from the home entity includes metadata from one or more data producing objects of the home entity. The method includes accessing action conditions associated with a user account. The action conditions identify a position where at least one or more states of the metadata from each of the home entity and the vehicle entity intersect. And, each action condition identifies a type or types of control information to be processed. The method includes processing the received metadata from the vehicle entity and the home entity.

Abstract: Digital content is received and supplemental content metadata is produced. The supplemental content metadata indicates a location of a feature in the digital content that is predicted to be of interest to a user. A digital content package is created that includes the digital content and the supplemental content metadata. The digital content package is provided to an electronic device, which presents the digital content in conjunction with a notification that a current position in the digital content is approaching the location of the feature.

Abstract: A predictive modeling method may include obtaining a fitted, first-order predictive model configured to predict values of output variables based on values of first input variables; and performing a second-order modeling procedure on the fitted, first-order model, which may include: generating input data including observations including observed values of second input variables and predicted values of the output variables; generating training data and testing data from the input data; generating a fitted second-order model of the fitted first-order model by fitting a second-order model to the training data; and testing the fitted, second-order model of the first-order model on the testing data. Each observation of the input data may be generated by (1) obtaining observed values of the second input variables, and (2) applying the first-order predictive model to corresponding observed values of the first input variables to generate the predicted values of the output variables.

Abstract: The present principles are directed to identifying and classifying web traffic inside encrypted network tunnels. A method includes analyzing network traffic of unencrypted data packets to detect packet traffic, timing, and size patterns. The detected packet, timing, and size traffic patterns are correlated to at least a packet destination and a packet source of the unencrypted data packets to create at least one of a training corpus and a model built from the training corpus. The at least one of the corpus and model is stored in a memory device. Packet traffic, timing, and size patterns of encrypted data packets are observed. The observed packet traffic, timing, and size patterns of the encrypted data packets are compared to at least one of the training corpus and the model to classify the encrypted data packets with respect to at least one of a predicted network host and predicted path information.

Abstract: A score explanation method for explaining a score includes at least steps of: a1) providing a first score associated with a first vector containing the first values of the parameters; b) generating a first set of lists, each list including a second number of indicators; c) generating, from a list, of at least a third vector wherein each parameter has a third value; the third value being equal to the corresponding first value when the list does not include an indicator of the corresponding parameter, and different from the corresponding first value otherwise; d) calculating the score of at least one third vector; e) evaluating, from the scores calculated for each of the third vectors, an indicator of significance of each parameter; f) elaborating, from the evaluated indicators of significance, an explanation of the first score.

Abstract: A system, a machine-readable storage medium storing instructions, and a computer-implemented method are described herein are directed to a Candidate Engine. The Candidate Engine generates a key(s) based on respective shared attributes between first profile data of a first member account and second profile data of a second member account in a social network service, The Candidate Engine assembles, according to encoded rules of a prediction model, feature vector data for each key. The encoded rules comprises at least one pre-defined feature predictive of an affinity between the first member account and the second member account. The Candidate Engine processes, according to the prediction model, the feature vector data for each key. The Candidate Engine receives predictive output from the prediction model.

Abstract: A heterogeneous log pattern editing recommendation system and computer-implemented method are provided. The system has a processor configured to identify, from heterogeneous logs, patterns including variable fields and constant fields. The processor is also configured to extract a category feature, a cardinality feature, and a before-after n-gram feature by tokenizing the variable fields in the identified patterns. The processor is additionally configured to generate target similarity scores between target fields to be potentially edited and other fields from among the variable fields in the heterogeneous logs using pattern editing operations based on the extracted category feature, the extracted cardinality feature, and the extracted before-after n-gram feature. The processor is further configured to recommend, to a user, log pattern edits for at least one of the target fields based on the target similarity scores between the target fields in the heterogeneous logs.