Abstract: An image similarity determination apparatus includes a memory and a processor configured to acquire a first image and a second image, perform selection of a first group and a second group from a plurality of feature points included in the first image and perform selection of a third group and a fourth group from a plurality of feature points included in the second image, calculate feature quantity for each feature point included in the first group and the third group on the basis of luminance and calculate feature quantity for each feature point included in the second group and the fourth group on the basis of hue, and determine similarity between the first image and the second image on the basis of both first comparison of the first group with the third group and second comparison of the second group with the fourth group.

Abstract: A method for learning parameters of an object detector based on a CNN adaptable to customer's requirements such as KPI by using a target object estimating network and a target object merging network is provided. The CNN can be redesigned when scales of objects change as a focal length or a resolution changes depending on the KPI. The method includes steps of: a learning device instructing convolutional layers to generate a k-th feature map by applying convolution operations to a k-th manipulated image which corresponds to the (k?1)-th target region on an image; and instructing the target object merging network to merge a first to an n-th object detection information, outputted from an FC layer, and backpropagating losses generated by referring to merged object detection information and its corresponding GT. The method can be useful for multi-camera, SVM (surround view monitor), and the like, as accuracy of 2D bounding boxes improves.

Abstract: Technologies are provided for automated crime scene analysis using machine learning. Firearm models, types, or even specific firearms may be automatically detected from captured audio files or continuous audio streams (e.g., recording microphones) using machine learning techniques. The detection may also be based on (or enhanced by) captured still images or video files/streams. Further information such as crime scene layout, wound types and locations, and similar information may be provided to the analysis service through manual input or automated capture (e.g., through analysis of image/video data). A number of firearms used in the commission of the crime may also be detected. Specific firearm types may be associated with specific crime types. Similar techniques may also be used to detect and classify types and quantity of explosive material.

Abstract: A method and an apparatus for recognizing characters using an image are provided. A camera is activated according to a character recognition request and a preview mode is set for displaying an image photographed through the camera in real time. An auto focus of the camera is controlled and an image having a predetermined level of clarity is obtained for character recognition from the images obtained in the preview mode. The image for character recognition is character-recognition-processed so as to extract recognition result data. A final recognition character row is drawn that excludes non-character data from the recognition result data. A first word is combined including at least one character of the final recognition character row and a predetermined maximum number of characters. A dictionary database that stores dictionary information on various languages using the first word is searched, so as to provide the user with the corresponding word.

Abstract: A radio frequency identification tag at a plurality of detection locations is detected and an expected location of the radio frequency identification tag is determined. A confidence level for each of the detection locations that detected the radio frequency identification tag is determined based on a relative location of the corresponding detection location as compared to the expected location. The determined confidence levels of the detection locations is analyzed to select at least one of the detection locations and an action based on the at least one selected detection location is performed.

Abstract: Techniques to improve detection and security of images, including formation and detection of matrix-based images. Some techniques include logic to process image data, generate one or more colorspaces associated with that data, and perform colorspace conversions based on the generated colorspace. The logic may be further configured to generate an image based on the colorspace conversions, including but not limited to a matrix bar code. The logic may be further configured to apply one or both of an ultraviolet layer and an infrared layer to the image, e.g. matrix barcode, generated from the colorspace conversion(s). Other embodiments are described and claimed.

Abstract: The invention relates to an electronic card comprising a card body (1) comprising an upper surface (11) and a lower surface (12); a fingerprint sensor (2) flush with the upper surface of said card body; a tactile mark (3) placed on the upper surface of the card body near the fingerprint sensor, said tactile mark having an excess thickness from the upper surface of the card to guide, when said mark is touched, a correct positioning of a finger on said fingerprint sensor, the thickness of the tactile mark being such that the card has a thickness compatible with standard ISO 781 ID1, the thickness of the tactile mark preferably being between 30 and 60 ?m, typically 50 ?m.

Abstract: A method for providing identification and access with respect to a passive radio-frequency tag in a tag population, the passive tag being configured for receiving ACK commands including an identification parameter, the method including receiving an initiating command for identifying and accessing a tag, the initiating command including a slot number having a first value if a value in a slot counter of the tag is equal to the first value, then transitioning to an Open or a Secured state, otherwise transitioning to an Arbitrate state in a case where the tag has transitioned to the Arbitrate state, receiving a command for repeating a tag access request, referred to as repetition command, the command including a slot number having a second value, different from the first value, the second value being equal to the value in the slot counter of the tag, transitioning to the Open or Secured state, receiving an ACK command, ignoring the ACK command.

Abstract: A near field communication (NFC) ring and method of making a ring capable of NFC, the method including the steps of: providing an electrically conductive band with two ends separated by a through slit, fitting an NFC coil having an NFC antenna and NFC chip against a sleeve that is smaller in circumference than the NFC coil and is not electrically conductive; and attaching the sleeve to the inside of the band, thereby forming an NFC enabled ring.

Abstract: A state machine engine having a program buffer. The program buffer is configured to receive configuration data via a bus interface for configuring a state machine lattice. The state machine engine also includes a repair map buffer configured to provide repair map data to an external device via the bus interface. The state machine lattice includes multiple programmable elements. Each programmable element includes multiple memory cells configured to analyze data and to output a result of the analysis.

Abstract: The present disclosure is drawn to the reduction of parameters in fully connected layers of neural networks. For a layer whose output is defined by y=Wx, where y is the output vector, x is the input vector, and W is a matrix of connection parameters, vectors uij and vij are defined and submatrices Wi,j are computed as the outer product of uij and vij, so that Wi,j=vij?uij, and W is obtained by appending submatrices Wi,j.

Abstract: Disclosed herein are systems and methods that identify and improve upon false positive scores. Some embodiments may include artificial neural network learning methods that utilize data input from users as well as enterprise machines. Information, such as transaction timing, prior transaction data, and demographics, may be taken as inputs to the neural network processing. The processing may be done on one or more neural network computers or nodes.

Abstract: Aspects of a neural network operation device are described herein. The aspects may include a matrix element storage module configured to receive a first matrix that includes one or more first values, each of the first values being represented in a sequence that includes one or more bits. The matrix element storage module may be further configured to respectively store the one or more bits in one or more storage spaces in accordance with positions of the bits in the sequence. The aspects may further include a numeric operation module configured to calculate an intermediate result for each storage space based on one or more second values in a second matrix and an accumulation module configured to sum the intermediate results to generate an output value.

Abstract: A neuromorphic device may include: a pre-synaptic neuron; a plurality of post-synaptic neurons; and a plurality of synapses electrically connected to the pre-synaptic neuron and electrically connected to the plurality of post-synaptic neurons. Each of the post-synaptic neurons may include: an integrator; a main comparator having a first input port connected to an output port of the integrator; and a first sub comparator having a first input port connected to the output port of the integrator.

Abstract: A system and method of detecting an aberrant message is provided. An ordered set of words within the message is detected. The set of words found within the message is linked to a corresponding set of expected words, the set of expected words having semantic attributes. A set of grammatical structures represented in the message is detected, based on the ordered set of words and the semantic attributes of the corresponding set of expected words. A cognitive noise vector comprising a quantitative measure of a deviation between grammatical structures represented in the message and an expected measure of grammatical structures for a message of the type is then determined. The cognitive noise vector may be processed by higher levels of the neural network and/or an external processor.

Abstract: This invention solves the long-standing problem in Machine Learning of training a neural network on a spike-based neuromorphic computer. The preferred embodiment of the invention describes an algorithm for training a Restricted Boltzmann Machine (RBM) neural network, but the invention applies equally to training neural networks in the general class of Markov Random Fields. The standard CD algorithm for training an RBM on a general-purpose computer is unsuitable for implementation on a neuromorphic computer, as it requires the communication of real-valued parameter values between neurons, and/or shared memory access by neurons to stored parameter values. By employing the invention described, these requirements are eliminated, thus providing a training algorithm which can be implemented efficiently on a spike-based, distributed processor and memory, neuromorphic computer system.

Abstract: Aspects described herein may allow for the application of stochastic gradient boosting techniques to the training of deep neural networks by disallowing gradient back propagation from examples that are correctly classified by the neural network model while still keeping correctly classified examples in the gradient averaging. Removing the gradient contribution from correctly classified examples may regularize the deep neural network and prevent the model from overfitting. Further aspects described herein may provide for scheduled boosting during the training of the deep neural network model conditioned on a mini-batch accuracy and/or a number of training iterations. The model training process may start un-boosted, using maximum likelihood objectives or another first loss function.

Abstract: Aspects described herein may allow for the application of stochastic gradient boosting techniques to the training of deep neural networks by disallowing gradient back propagation from examples that are correctly classified by the neural network model while still keeping correctly classified examples in the gradient averaging. Removing the gradient contribution from correctly classified examples may regularize the deep neural network and prevent the model from overfitting. Further aspects described herein may provide for scheduled boosting during the training of the deep neural network model conditioned on a mini-batch accuracy and/or a number of training iterations. The model training process may start un-boosted, using maximum likelihood objectives or another first loss function.

Abstract: A speech recognition neural network system includes an encoder neural network and a decoder neural network. The encoder neural network generates an encoded sequence from an input acoustic sequence that represents an utterance. The input acoustic sequence includes a respective acoustic feature representation at each of a plurality of input time steps, the encoded sequence includes a respective encoded representation at each of a plurality of time reduced time steps, and the number of time reduced time steps is less than the number of input time steps. The encoder neural network includes a time reduction subnetwork, a convolutional LSTM subnetwork, and a network in network subnetwork. The decoder neural network receives the encoded sequence and processes the encoded sequence to generate, for each position in an output sequence order, a set of substring scores that includes a respective substring score for each substring in a set of substrings.

Abstract: The prediction function creation device according to the present invention for creating a prediction function to derive an objective variable by using a set of samples that include explanatory variables and an objective variable, the device includes: a clustering unit that clusters the respective samples by giving labels, and assigns weights to each label in accordance with patterns of missing values for the explanatory variables in labeled samples; a child model creation unit that makes portions of the training data partial training data on the basis of the weights, and determines an explanatory variable that constitutes the prediction function on the basis of patterns of missing values for the explanatory variables in the samples; and a mixture model creation unit that creates the prediction function with respect to each pattern of missing values by using the explanatory variable and the determined partial training data.

Abstract: Disclosed herein is a computer architecture and software that is configured to modify handling of predictive models by an asset-monitoring system based on a location of an asset. In accordance with example embodiments, the asset-monitoring system may maintain data indicative of a location of interest that represents a location in which operating data from assets should be disregarded. The asset-monitoring system may determine whether an asset is within the location of interest. If so, the asset-monitoring system may disregard operating data for the asset when handling a predictive model related to the operation of the asset.

Abstract: Systems and methods for generating performance prediction model and estimating execution time for applications is provided. The system executes synthetic benchmarks for a first dataset on a first cluster. Each synthetic benchmark includes a MapReduce (MR) job. The system further extracts sensitive parameters for each sub-phase of the MR job, generates a linear regression prediction model for each sub-phase to obtain one or more linear regression prediction models, based on which the system further generates a performance prediction model to be utilized for predicting, using the sensitive parameters, a Hive query execution time of a Directed Acyclic Graph (DAG) of one or more MR jobs executed on a second dataset on a second cluster, wherein the first cluster that includes the first dataset is smaller compared to the second cluster that includes the second dataset.

Abstract: Group-based information displays correspond to the collection of display objects representative of either contacts associated with a user or applications or executable components executable, or otherwise accessible, via the mobile communication device. The group-based information displays can be logically organized according to subject matter organizational criteria. The subject matter organizational criteria can include, for example, organization or other affiliates criteria, event-based, or activity based criteria, topic-based criteria and the like. Through the group-based information displays, mobile communication device users can initiate one or more actions corresponding to the selection of display objects corresponding to applications or executable components. Additionally, as applicable, the user can limit, or otherwise apply, the initiated actions to contacts identified, or otherwise associated, with the specific group-based information display.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training and applying a machine learning model. One of the methods includes the actions of obtaining a collection of training data, the training data comprising collection of data points associated with a labeled set of real property parcels; training a machine learning model using the training data, the machine learning model being trained to generate a likelihood with respect to a parameter from input data associated with a specific parcel of real property, wherein training includes optimizing the model using a Markov chain optimization that seeks to minimize error in the model where the model is underpinned by one or more non-differentiable functions; receiving a plurality of data points associated with an input parcel of real property; and using the optimized model to generate a likelihood for the parameter for the input parcel of real property.

Abstract: This invention comprises a method of simulating an ecological environment, where digital agents within the environment are capable of processing data, and agents that successfully process data are permitted to reproduce to generate new algorithms. The invention is a groundbreaking advance in artificial intelligence and machine learning and enables processes that were once considered computationally impossible.

Abstract: An efficient fact checking system analyzes and determines the factual accuracy of information and/or characterizes the information by comparing the information with source information. The efficient fact checking system automatically monitors information, processes the information, fact checks the information efficiently and/or provides a status of the information.

Abstract: Providing predictive data predicting data values for a historical dataset. The method facilitates improving the accuracy of the predictive data by identifying for a user, and allowing the user to select ancillary datasets that can be evaluated, using a predictive evaluation, together with a historical dataset to improve the accuracy of the predictive data. A user interface is provided to a user. The user interface identifies one or more ancillary datasets. The ancillary datasets are identified to the user based on the ancillary datasets meeting a threshold condition to a historical dataset. The ancillary datasets are selectable by the user in the user interface. User input is received at the user interface selecting one or more of the ancillary datasets. A predictive dataset is displayed to the user. The predictive dataset is determined by predictive evaluation of the historical dataset and the one or more selected ancillary datasets.

Abstract: In implementations of the subject matter described herein, each token for containing an element in the training data is sampled according to a factorization strategy in training. Instead of using a single proposal, the property value of the target element located at the token being scanned is iteratively updated one or more times based on a combination of an element proposal and a context proposal. The element proposal tends to accept a value that is popular for the target element independently of the current piece of data, while the context proposal tends to accept whenever the property value that is popular in the context of the target data or popular for the element itself. The proposed modeling training approach can converge in a quite efficient way.

Abstract: The disclosed embodiments provide a system for processing user actions with a service. During operation, the system uses a statistical model to obtain a first metric associated with a user action received by a service. Next, the system applies a set of static decisions to the metric and one or more attributes of the user action to determine a first response to the user action. The system then uses a set of dynamic rules to produce a first modification to the first response. Finally, the system generates output for applying the first response to the user action.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for constructing and programming quantum hardware for quantum annealing processes.

Abstract: A composition of autonomous self-learning skill software components interact with one another to solve a real time complex task. Each software component includes a knowledge base and an inference algorithm that uses the knowledge base to solve a given data-processing task involving input data. Each software component may also include a machine learning algorithm for training the knowledge base with new data in real-time as the new data is received.

Abstract: In one embodiment, a processor of a computing device receives a query. The computing device may compare a centroid of each of a plurality of clusters to the query such that a subset of the plurality of clusters is selected, each of the plurality of clusters having a set of data points. An assignment of the subset of the plurality of clusters may be communicated to a hardware accelerator of the computing device. A plurality of threads of the hardware accelerator of the computing device may generate one or more distance tables that store results of intermediate computations corresponding to the query and the subset of the plurality of clusters. The distance tables may be stored in shared memory of the hardware accelerator. A plurality of threads of the hardware accelerator may determine a plurality of data points using the distance tables. The processor may provide query results pertaining to at least a portion of the plurality of data points.

Abstract: A stream of attention method, system, and apparatus determines a first focus of attention by applying a probabilistic analysis to a first set of syntactical elements. The syntactical elements may be derived from external communications or from communications generated internally by a computer-implemented system. Alternatively, the syntactical elements may be determined based on their association with images accessed from an external or internal source. Actions are performed by the computer-implemented system that are expected to reduce uncertainty with respect to the focus of attention. Such actions may include posing syntactical-based interrogatives directed to external entities or internally, searching a corpus of content, and/or moving to a new location and receiving input at the new location.

Abstract: Techniques for deriving additional features from input data are described herein. Input data from a plurality of source files are received. One or more features corresponding to the input data, which includes information about semantic types, is identified. The input data is then processed to generate additional features for the input data. New data corresponding to the additional features are then generated and access to the new data is subsequently provided.

Abstract: Systems and methods for evaluating a loss function or a gradient of the loss function. In one example embodiment, a computer-implemented method includes partitioning a weight matrix into a plurality of blocks. The method includes identifying a first set of labels for each of the plurality of blocks with a score greater than a first threshold value. The method includes constructing a sparse approximation of a scoring vector for each of the plurality of blocks based on the first set of labels. The method includes determining a correction value for each sparse approximation of the scoring vector. The method includes determining an approximation of a loss or a gradient of a loss associated with the scoring function based on each sparse approximation of the scoring vector and the correction value associated with the sparse approximation of the scoring vector.

Abstract: Systems and methods for machine learning, models, and related explainability and interpretability are provided. A computing device determines a contribution of a feature to a predicted value. A feature computation dataset is defined based on a selected next selection vector. A prediction value is computed for each observation vector included in the feature computation dataset using a trained predictive model. An expected value is computed for the selected next selection vector based on the prediction values. The feature computation dataset is at least a partial copy of a training dataset with each variable value replaced in each observation vector included in the feature computation dataset based on the selected next selection vector. Each replaced variable value is replaced with a value included in a predefined query for a respective variable. A Shapley estimate value is computed for each variable.

Abstract: Product providers can offer matched prices for the products to the customers through a platform surveying the available prices from sale agents. The platform can determine lowest prices from the sale agents, and use this information to have an agreement with the product providers to sell the products at compatible prices.

Abstract: Systems, methods, and computer program products for coordinating operation of relevant systems after a passenger itinerary disruption. Responsive to a transfer request being received that includes at least a portion of a new travel itinerary to replace a disrupted travel itinerary of each of multiple passengers, an inventory system automatically updates counters based on the disrupted travel itinerary of each passenger and the new travel itinerary, and a reservation system automatically updates one or more reservation records to reflect an association between each passenger and the new travel itinerary. Thereafter, a ticketing system automatically performs a ticketing process for each passenger relative to the new travel itinerary, and a departure control system automatically transfers passenger data stored for each passenger in association with the disrupted travel itinerary of the passenger to a record associated with the new travel itinerary.

Abstract: A computer-implemented method includes receiving a new data record associated with a transaction, and generating, using an adaptive model executed by the computer, a score to represent a likelihood that the transaction is associated with fraud. The adaptive model employs feedback from one or more external data sources, the feedback containing information about one or more previous data records associated with fraud and non-fraud by at least one of the one or more external data sources. Further, the adaptive model uses the information about the one or more previous data records as input variables to update scoring parameters used to generate the score for the new data record.

Abstract: This disclosure describes, according to some implementations, a method for scoring electronic meeting requests for attendee redundancy. In an example method, the method includes retrieving, using one or more processors, user IDs of meeting attendees; retrieving, using the one or more processors, one or more characteristics associated with the user IDs; retrieving, using the one or more processors, a rule having one or more parameters for scoring meeting composition; comparing, using the one or more processors, the one or more characteristics associated with the user IDs based on the one or more parameters of the rule; and generating, using the one or more processors, a meeting score based on the comparison.

Abstract: A cloud-based global alarm annunciation broker assist in locating and contacting suitable technical support personnel in response to detected alarm events at a plant facility. The system comprises a cloud-based framework that dynamically matches on-site alarm events to domain experts capable of addressing the alarm events. The framework uses an agent-based architecture to gather industrial data from data sources within the industrial environment, including time-series alarm data. The data is received at a cloud platform, where broker services perform a global search for suitable technical support resources in response to alarm events identified by the collected data. The brokering system can automatically generate service tickets and send notifications to both end users and system managers/supervisors, and can notify application-level experts about events and anomalies that emerge from the on-premise processes being monitored by the cloud-based broker system.

Abstract: A method, apparatus and computer program product are provided in order to utilize task value units for imaging interpretation and other tasks, such as in the assignment of imaging interpretation and other tasks to a plurality of users. In the context of a method, the method associates, for each of a plurality of different types of imaging interpretation tasks and for each of a plurality of other tasks, a first task value unit therewith. The method also associates, for at least some of the plurality of different types of imaging interpretation or other tasks, a second task value unit, different than the first task value unit, therewith. Further, the method assigns imaging interpretation and other tasks to a plurality of users based at least partially upon the first and second task value units associated with the respective imaging interpretation tasks and the other tasks.

Abstract: A system tool that provides dispatchers in power grid control centers with a capability to manage changes. Included is a user interface and a plurality of scheduler engines. Each scheduler engine is configured to look ahead at different time frames to forecast system conditions and alter generation patterns within the different time frames. A comprehensive operating plan holds schedules generated by the plurality of scheduler engines. A relational database is coupled to the comprehensive operating plan. Input data is initially received from the relational database for each scheduling engine, and thereafter the relational database receives data from the scheduling engines relative to forecast system conditions.

Abstract: Techniques for modeling a data center are provided. In one aspect, a method for determining data center efficiency is provided. The method includes the following steps. Target parameters for the data center are obtained. Technology pre-requisite parameters for the data center are obtained. An optimum data center efficiency is determined given the target parameters for the data center and the technology pre-requisite parameters for the data center.

Abstract: In various embodiments, a system may be configured to substantially automatically determine whether to take one or more actions in response to one or more identified risk triggers (e.g., data breaches, regulation change, etc.). The system may, for example: (1) compare the potential risk trigger to one or more previous risks triggers experienced by the particular entity at a previous time; (2) identify a similar previous risk trigger (e.g., one or more previous risk triggers related to a similar change in regulation, breach of data, type of issue identified, etc.); (3) determine the relevance of the current risk trigger based at least in part on a determined relevance of the previous risk trigger; and (4) determine whether to take one or more actions to the current risk trigger based at least in part on one or more determined actions to take in response to the previous, similar risk trigger.

Abstract: A data collection method including collecting, by a configuration collector manager, configuration data, including configuration properties, from a plurality of data sources, creating, by a model discovery component, a business application model using the configuration data collected by the configuration collector manager, a collecting, from a business application model analysis user interface, edits and confirmations associated with the business application model, and a analyzing, by a feedback analyzer component, the edits and confirmations associated with the business application model and prioritizing the configuration properties based on the data sources and the edits and confirmations associated with the business application model.

Abstract: Even with modern high-speed processors and bandwidth, many systems implement intensive data reporting components that tax resources. Even very small improvements in reporting efficiencies can result in significant cost and resource savings, as well as improvements in the performance of such resources. Recording facts (e.g., events) with a standardized time and time interval based upon a universal reference, such as Coordinated Universal Time (UTC), and reporting based upon local time and local intervals improves data processing throughput with limited report granularity. For example, since forty-eight intervals comprise most of the world's time zones, reports simply designating facts as occurring within one of the forty-eight intervals and one of the thirty-minute intervals greatly improves operability and throughput.

Abstract: A compliance determination and enforcement platform is described. A plurality of factors are stored in association with each of a plurality of accounts. A factor entering module enters factors from each user account into a compliance score model. The compliance score model determines a compliance score for each one of the accounts based on the respective factors associated with the respective account. A comparator compares the compliance score for each account with a compliance reference score to determine a subset of the accounts that fail compliance and a subset of the accounts that meet compliance. A flagging unit flags the user accounts that fail compliance to indicate non-compliant accounts. A corrective action system allows for determining, for each one of the accounts that is flagged as non-compliant, whether the account is bad or good, entering the determination into a feedback system and closing the account.

Abstract: This patent specification relates to apparatus, systems, methods, and related computer program products for providing home security/smart home objectives. More particularly, this patent specification relates to a plurality of devices, including intelligent, multi-sensing, network-connected devices, that communicate with each other and/or with a central server or a cloud-computing system to provide any of a variety of useful home security/smart home objectives.

Abstract: A package delivery apparatus uses an unmanned aerial vehicle (UAV) to deliver a package containing a product to a delivery destination area. The UAV uses GPS signals to guide it to the delivery destination area and an altimeter to determine its height above the delivery destination area. The UAV then adjusts its height to a preferred drop or release height for that package and product and releases the package. An optional camera allows a human operator to view the delivery destination area. An expandable foam package surrounds the product to protect the product from impact and moisture. The package may be streamlined to reduce air resistance and increase the range of the UAV. The package characteristics, such as its thickness, are determined based one or more of the weight and fragile nature of the product, and the drop height.