Abstract: Systems and methods involving integrating camera and acoustic sensor data, and automatically capturing the acoustic sensor heatmap for the holistic sensing systems in Internet of Things (IoT) systems. In particular, example implementations described herein capture the local sound noise environment or localized noise profiles (e.g., noise fingerprint) adaptively to the change of noise profiles and automatically apply captured noise profiles to the streaming noise reduction in signal processing for industrial IoT areas.

Abstract: Application performance data and machine health are collected by a system. The system correlates the two data types to provide context as to how machine health affects the performance of an application. Performance data for an application, for example an application executing as part of a distributed business transaction, and health data for a machine which hosts the application are collected. The performance data and machine health data may be correlated for a particular period of time. The correlation may then be reported to a user. By viewing the correlation, a user may see when machine health was good and bad, and may identify the effects of the machine health on the performance of an application.

Abstract: A data portion of a data set utilized in a computerized visualization is analyzed to identify one or more areas of interest each including data values representing distinguishable features relative to the data set. An explanation for the data values of each of the one or more areas of interest is determined. Each explanation is based on other data portions of the data set contributing to the distinguishable features. At least one display layer including labels describing the one or more areas of interest is generated. The labels include the explanation for each of the one or more areas of interest. The at least one display layer is disposed over the computerized visualization to produce an annotated visualization with the labels positioned proximate the one or more areas of interest.

Abstract: This disclosure relates generally to methods and systems for determining whether at least a first component and a second component of a personal protection (PP) configuration are compatible so as to be assembled into an approved PP configuration, responsive to being compared against at least one predetermined criterion. Included is a smart tag for each of the first and second components. The first and second smart tags include respective first and second data. A data acquiring system is provided for acquiring the first and second data and a data processing system processes the acquired first and second data for determining if the first and second components are compatible to be assembled as an approved PP configuration.

Abstract: Methods and systems for inline parts average testing and latent reliability defect recognition or detection are disclosed. An inline parts average testing method may include: performing inline inspection and metrology on a plurality of wafers at a plurality of critical steps during wafer fabrication; aggregating inspection results obtained from inline inspection and metrology utilizing one or more processors to obtain a plurality of aggregated inspection results for the plurality of wafers; identifying one or more statistical outliers among the plurality of wafers at least partially based on the plurality of aggregated inspection results obtained for the plurality of wafers; and disqualifying the one or more statistical outliers from entering a supply chain for a downstream manufacturing process, or segregating the one or more statistical outliers for further evaluation, testing or repurposing.

Abstract: Anomalous sensors are detected using an apparatus including a processor and one or more computer readable mediums collectively including instructions that, when executed by the processor, cause the processor to: obtain a plurality of healthy sensor data, wherein each of the healthy sensor data includes a plurality of sensed values of a corresponding sensor among a plurality of sensors in normal operation, generate a healthy data distribution of at least two sensors among the plurality of sensors based on the plurality of healthy sensor data, and generate a function of a status probability distribution of the plurality of sensors with respect to time under the condition of sensor data with respect to time based on the healthy data distribution.

Abstract: A system for performing sensor maintenance, including redundant sensors to collect one or more data points; one or more non-transitory storage mediums; a processor; and an output device; wherein one of the non-transitory storage mediums receives the one or more data points; wherein one of the non-transitory storage mediums includes one or more instructions for performing two or more statistical calculations; and wherein the processor carries out the instructions. The method for using the system includes: the non-transitory storage medium receiving one or more data points from the redundant sensors; the processor performing the statistical calculations on the data points to generate one or more processed data points and comparing the processed data points to a threshold value; and the non-transitory storage medium creating an alert when the processed data points exceed a threshold value for a period of time; and displaying the alert on the output device.

Abstract: A method for deploying application components to Information Technology (IT) resources using an application platform controller includes requesting a network slice for the application components from a network controller. Priority values are assigned to the application components defining an order in which the application components are planned to be deployed. The application components are deployed sequentially, using the priority values, to host sites of the IT resources using information from the network controller including existing registrations of users of the application to an access network of the host sites such that replicas of the application components are provided at the host sites which provide a minimum latency with respect to a location of the users.

Abstract: A system may involve persistent storage containing a configuration management database (CMDB) and a non-CMDB table, wherein the CMDB contains configuration items that represent software, devices, or services deployed within a network, and wherein the non-CMDB table contains entries related to operation of the network. The system may also involve one or more processors configured to provide a representation of a graphical user interface (GUI), wherein the GUI contains a first selectable tab that displays classes of the configuration items, a second selectable tab that displays the non-CMDB table, and a canvas for visually depicting query expressions, wherein the classes are selectable to place class GUI elements thereof onto the canvas, wherein the non-CMDB table is selectable to place a table GUI element thereof onto the canvas, and wherein the table GUI element and a particular class GUI element are connectable by a link on the canvas.

Abstract: A method for estimating consumed battery life of at least one battery of a pitch drive mechanism of a rotor blade of a wind turbine includes monitoring, via at least one sensor, an actual temperature of the battery over a predetermined time period. The method also includes storing, via a turbine controller, the monitored actual temperatures of the battery during the predetermined time period. Further, the method includes determining, via the turbine controller, the consumed battery life as a function of the monitored actual temperatures.

Abstract: In some examples, a system includes a plurality of articles of personal protected equipment (PPE) that are each assigned to a particular worker. The system may also include a data hub that detects an input that initiates a broadcast of diagnostic self-check messages; identifies, in response to the input, each article of PPE of the plurality of articles of PPE; broadcasts, based on identifying each article of PPE, the diagnostic self-check messages to the respective articles of PPE, wherein each article of PPE receives its respective self-check message at its communication component; in response to receiving a set of diagnostic acknowledgement messages from one or more of the plurality of articles of PPE that have performed a diagnostic self-check, determines whether the set of diagnostic acknowledge messages satisfy one or more self-check criteria; and performs one or more operations based on whether the self-check criteria are satisfied.

Abstract: A system and method for determining a remaining useful life of a component, includes providing a usage and configuration information associated with the component, determining if a preexisting spall condition exists for the component, executing a spall initiation model for the component if the preexisting spall condition does not exist, wherein the spall initiation model provides a spall initiation life associated with the component by analyzing the usage and configuration information, executing a spall propagation model for the component, wherein the spall propagation model provides a spall propagation life associated with the component by analyzing the usage and configuration information, and providing a remaining useful life by integrating the spall initiation life and the spall propagation life.

Abstract: An embodiment of the invention may include a method, computer program product and system for guided service procedure. The embodiment may include receiving feedback data from one or more of a plurality of sensors. Each of the plurality of sensors may detect a physical condition at a serviceable location within an item of equipment. The item of equipment may be undergoing a service procedure by an equipment servicer. The embodiment may include determining whether the received feedback data from one or more of the plurality of sensors exceeds a threshold value. Based on determining that the threshold value is exceeded, the embodiment may include alerting the equipment servicer during the service procedure.

Abstract: In embodiments, a mechanical solar tracker disclosed herein includes a frame, an elevation cam with an elevation cam surface attached to the frame, and a rotation drive mechanically coupled to the frame. A collector carriage is coupled to the rotation drive and configured to rotate circumferentially around a vertical axis. Also, a collector is connected to the collector carriage and a hinge is coupled between the collector and the collector carriage wherein the hinge is configured to rotate the collector about a horizontal plane at a pivot point. Additionally, a cam follower is coupled between the collector and the elevation cam and the cam follower is configured to translate, based on a surface input from the elevation cam surface, rotation of the collector about the pivot point to track the sun's direction, by tracing the elevation cam surface with the cam follower.

Abstract: A method for state monitoring a bearing that supports a planetary gear of a planetary transmission on a planet carrier, the method including measuring a structure-borne noise of the planetary transmission at a stationary position on the planetary transmission in order to generate a digital measurement signal having a sample value at at least one sample time point, determining a relative speed of the planetary gear with respect to the stationary position for the sampling time point, determining a corrected measurement signal, the corrected measurement signal including the sample value with a corrected sample time point, the corrected sample time point being shifted with respect to the sample time point based on the relative speed, forming an envelope-curve signal based on the corrected measurement signal, determining a frequency spectrum for the envelope-curve signal, and comparing the frequency spectrum with a reference spectrum in a given frequency range.

Abstract: Embodiments are directed to systems that build and deploy inferential models for generating predictions of a plant process. The systems select input variables and an output variable for the plant process. The systems load continuous measurements for the selected input variables. For the selected output variable, the systems load measurements of type: continuous from the subject plant process, intermittent from an online analyzer, or intermittent from lab data. If continuous or analyzer measurements are loaded, the systems build a FIR model with a subspace ID technique using continuous output measurements. From intermittent analyzer measurements, the systems generate continuous output measurements using interpolation. If lab data is loaded, the systems build a hybrid FIR model with subspace ID and PLS techniques, using continuous measurements of a reference variable correlated to the selected output variable.

Abstract: In one embodiment, a test system comprises a first interface for communicating with remote devices; a second interface for communicating with local devices; a memory for storing information, including information received from the first interface and second interface; a processor for automatically configuring test system components in accordance with the information stored in the memory. The test system components comprise a network access point simulation component and a local control component. The network access point simulation component is configured to simulate communication network access point operations comprising test interactions with user equipment. The number of devices under test included in the user equipment and distinct network access points that are coincidentally simulated can be variable. The local control component is configured to direct the network access point simulation component and to control the test interactions with the user equipment.

Abstract: A monitor device is configured to monitor a system, such as a utility system and/or appliance. The monitor device includes sensor(s) that measure characteristic(s) of the system, and data storage that stores parameter data that is provided from a remote service. The parameter data includes thresholds, signatures, and/or other information that describes a typical operational state of the system. The monitor device includes software that executes locally to compare current sensor data to the parameter data and determine whether the current operational state of the system is within the bounds of the typical state as defined by the parameter data. In instances where the current state is atypical, indicating a possible problem with the system, a notification can be sent describing the anomaly. The monitor device can also perform other action(s) to correct the problem or mitigate possible damage, and can operate autonomously if a network connection is unavailable.

Abstract: Embodiments of the invention include methods and systems for architectures for wireless avionics communication networks. The embodiments further include detecting a signal strength of wireless nodes, assigning a primary data controller and standby data controller for each of the wireless nodes based at least in part on the signal strength, and generating a deployment matrix based on the assignment of the primary data controller and the standby data controller. The embodiments also include broadcasting the deployment matrix over a wired connection, allocating a buffer size based on data rates of each of the wireless nodes connected to the primary data controller and the standby data controller, and exchanging data based on the deployment matrix.

Abstract: A method for optimizing a fluid utilization facility. The method includes monitoring an operating state of a fluid utilization device and an operating state of a drain trap in a fluid utilization facility based on detection information obtained by detectors installed in various places in the fluid utilization facility. A running state of the fluid utilization facility is optimized based on a monitoring result.

Abstract: Systems, methods, and computer-readable media for implementing local and wide network suppression of query requests in zero-configuration networking. Disclosed are systems, methods, and computer-readable storage media for implementing suppression of query packets zero-configuration networking over local and wide networks by utilizing agents, application programming interfaces (API), and a controller. The suppression can be determined based on two time periods, a processing period and a suppression period.

Abstract: Methods and systems for implementing experimental trials on utility grids. The variation of grid parameters are coordinated to create periods of time and areas of space from within which the variations of grid parameters do not overlap, allowing sensor data within those periods of time and areas of space to be associated with particular variations in grid parameters. This associated data can in turn be used to improve models of sensor response and utility grid behavior.

Abstract: In one example of the disclosure, a set of scanned images is accessed. The scanned images are scans of distinct printouts of subject images produced utilizing a photo imaging plate. A set of defect maps is created by comparing the scanned images to reference data for the subject images. The set of defect maps are combined into an integrated defect map. A scratch defect on the PIP is identified utilizing the integrated defect map.

Abstract: An information processing device according to one aspect of the present invention includes a first acquirer configured to acquire measured data of a sensor, a second acquirer configured to acquire characteristic data of the sensor, the characteristic data having been acquired by maintaining the sensor, and a generator configured to generate teacher data in which the acquired characteristic data is associated as label information with the acquired measured data.

Abstract: Aspects of the present invention include a method, which includes receiving, by a processor, data from one or more communicatively coupled objects associated with a device. The method further includes providing, by the processor, context information to the received data. The method further includes mapping, by the processor, the context information associated with the received data to one or more nodes of a knowledge graph. The method further includes diagnosing, by the processor, a problem with the device, based on knowledge graph information.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying broken network connections. In one aspect, a system includes front-end server(s) that receive data specifying, for multiple different user interactions with one or more application links that link to a given application, presentation durations specifying how long application content linked to by the application link was presented following the multiple different user interactions with the application link(s). Back-end server(s) that communicate with the front end server(s) can classify each application link as broken or working based on application of a machine learning model to the presentation durations for the application link. The machine learning model can be generated using labeled training data. The back-end server(s) can generate and output an alert identifying an application link as a broken link based on the application link being classified as broken by the machine learning model.

Abstract: Remote sensing techniques are being increasingly used for periodic structural health monitoring of vast infrastructures. Conventionally, analysis of visual and other signals captured from sensing devices are used to diagnose faults. Such data collection and analysis is expensive in terms of both computational overheads as well as towards robotic maneuvering of data collection systems, such as a UAV. In accordance with the present disclosure, the data acquisition system is modeled as an intelligent situated agent that autonomously controls data gathering and analysis activities through a cognitive cycle of perception-recognition-action, in order to optimize the cost of efforts in identifying faults that may exist. Also, a reactive, economical planning algorithm around Qualitative Bayesian Network (QBN) that controls the sequence of data collection and analysis has been implemented.

Abstract: An apparatus for controlling an inverter includes a volatile, first storage part; a nonvolatile, second storage part; and a control part configured to store data related to the status of the inverter in the first storage part when the inverter is driven and configured to select some of the data stored in the first storage part depending on the type of failure event when a failure event occurs and store the selected data in the second storage part.

Abstract: There is provided the diagnostic device for a sensor which is arranged in an exhaust passage of an internal combustion engine and detects a particulate matter amount in exhaust, the diagnostic device including a time-rate-of-change calculation unit which calculates a time rate of change of the particulate matter amount detected by the sensor during a period in which a fuel injection amount of the internal combustion engine is equal to or less than a predetermined injection amount threshold, and an abnormality determination unit which determines an abnormality of the sensor based on the time rate of change of the particulate matter amount calculated by the time-rate-of-change calculation unit.

Abstract: A parameter-based multi-model blending method and system are described. The method includes selecting a parameter of interest among parameters estimated by each of a set of individual models, running the set of individual models with a range of inputs to obtain a range of estimates of the parameters from each of the set of individual models, and identifying, for each of the set of individual models, critical parameters among the parameters estimated, the critical parameters exhibiting a specified correlation with an error in estimation of the parameter of interest. For each subspace of combinations of the critical parameters, obtaining a parameter-based blended model is based on blending the set of individual models in accordance with the subspace of the critical parameters, the subspace defining a sub-range for each of the critical parameters.

Abstract: A parameter-based multi-model blending method and system are described. The method includes selecting a parameter of interest among parameters estimated by each of a set of individual models, running the set of individual models with a range of inputs to obtain a range of estimates of the parameters from each of the set of individual models, and identifying, for each of the set of individual models, critical parameters among the parameters estimated, the critical parameters exhibiting a specified correlation with an error in estimation of the parameter of interest. For each subspace of combinations of the critical parameters, obtaining a parameter-based blended model is based on blending the set of individual models in accordance with the subspace of the critical parameters, the subspace defining a sub-range for each of the critical parameters.

Abstract: A method and apparatus of monitoring computer devices operating on a network is disclosed. Computer devices are all different and require monitoring settings that are tailored to their specific requirements. One example of the present invention may include a method of monitoring at least one computer device operating on a network. The method may include receiving audit information representing attributes of the computer device and storing the audit information in memory. The method may also include comparing the audit information to a predefined monitor set of objects to be monitored. The method may further include creating a new monitor set based on the comparison of the audit information and the predefined monitor set. The new monitor set is different from the predefined monitor set and is generally used to monitor objects which are included in the audited device. The method may also include monitoring the at least one computer device based on the new monitor set.

Abstract: In an example embodiment, for each of a plurality of different titles in a social network structure, the title is mapped into a first vector having n coordinates, while kills are mapped into a second vector having n coordinates. The first and second vectors are stored in a deep representation data structure. One or more objective functions are applied to at least one combination of two or more of the vectors in the deep representation data structure. Then, an optimization test on each of the at least one combination is performed using a corresponding objective function output for each of the at least one combination of two or more of the vectors, and, for any combination that did not pass the optimization test, one or more coordinates for the vectors in the combination are altered so that the vectors in the combination become closer together within an n-dimensional space.

Abstract: This disclosure relates generally to methods and systems for determining whether at least a first component and a second component of a personal protection (PP) configuration are compatible so as to be assembled into an approved PP configuration, responsive to being compared against at least one predetermined criterion. Included is a smart tag for each of the first and second components. The first and second smart tags include respective first and second data. A data acquiring system is provided for acquiring the first and second data and a data processing system processes the acquired first and second data for determining if the first and second components are compatible to be assembled as an approved PP configuration.

Abstract: A method of identifying at least one malfunctioning nozzle in a digital printing press including a plurality of nozzles. The method includes the procedures of printing a design on a substrate, acquiring at least one image of the printed design and identifying at least one artifact in the acquired image. The method further includes the procedure of identifying the malfunctioning nozzle and classifying the at least one malfunctioning nozzle according to at least a portion of a combined pattern.

Abstract: Techniques are described for detecting anomalous behavior in program execution. In one example, a method includes logging occurrence of one or more key run time events during execution of a program. Each key run time event has a corresponding key run time event data structure associated with the program, and logging includes storing records associated with the key run time events, wherein each record is based on the key run time event data structure associated with the key run time event. The method further includes analyzing the records to determine if a current pattern of key run time events associated with the program during execution matches an expected pattern of key run time events and generating a security alert if the current pattern of key run time events does not match the expected pattern of key run time events for the program.

Abstract: A computer program product and system for updating operating parameters of a consumer appliance. Respective operating conditions are received regarding a plurality of consumer appliances that are communicatively coupled to a database. Historical data tracking at least one component in the plurality of consumer appliances based on the operating conditions is generated. A trend, based on the data, as well as a first set of optimal operating parameters for operating the component in a first consumer appliance of the plurality of consumer appliances is determined. The first set of updated operating parameters is transmitted to the first consumer appliance to effect a change in an operation of the component in the first consumer appliance.

Abstract: The difference in operational management service among a plurality of storage apparatuses is detected by an apparatus that provides at least one operational management service to the storage apparatuses of a plurality of user sites via a communication network. The apparatus classifies the storage apparatuses of the plurality of user sites on the basis of use statuses, and generate at least one group including the plurality of storage apparatuses; and a service difference extraction unit configured to detect, out of the plurality of storage apparatuses that constitute the group, a first storage device that is inferior in terms of a predetermined index and a second storage device that is superior to the first storage device in terms of the predetermined index, and extract a difference between a first operational management service applied to the first storage apparatus and a second operational management service applied to the second storage apparatus.

Abstract: A system for monitoring stress cycles includes memory storing a base value and an inflection value of a stress cycle and one or more processors coupled to the memory. The one or more processors are configured to receive a series of stress values from a stress sensor. For each stress value in the series of stress values, the one or more processors are configured to perform operations including performing a first comparison between the stress value and a previous stress value in the series of stress values, detecting an inflection in the series of stress values based on the first comparison, updating the base value and the inflection value in response to detecting the inflection, performing a second comparison between the stress value and the base value, determining whether the stress cycle is complete based on the second comparison, and recording the stress cycle in response to determining that the stress cycle is complete.

Abstract: A method for detecting an anomaly in sensor data generated in a substrate processing apparatus is disclosed herein. A plurality of data sets is received. A first data set from a first sensor and second data set from a second sensor are selected. The first second sensors are defined as a sensor pair. A reference correlation is generated by selecting a subset of values in each data set for each of the first and second data sets. A difference of remaining data correlation outside the subset of values in each data set to the reference correlation is normalized. The normalized data set is filtered to smooth the normalized difference to avoid isolated outliers with high chance of false positive candidates. One or more anomalies are identified. Process parameters of the substrate processing apparatus are adjusted, based on the one or more identified anomalies from the filtered data set.

Abstract: A vehicle health monitoring system including a rules generation module that receives historical time-stamped vehicle fault data for a vehicle system, determines a correlation between pairs of a precedent historical vehicle fault data, of the historical time-stamped vehicle fault data, and a subsequent different historical vehicle fault data, of the historical time-stamped vehicle fault data, and generates maintenance rules based on the correlation determined for the pairs. A fault detection module of the system monitors faults of the vehicle system, where the vehicle faults include a plurality of time-stamped precedent vehicle fault data, applies the rules to the plurality of time-stamped precedent vehicle fault data, determines an imminent occurrence of a subsequent vehicle fault, and generates a maintenance report corresponding to the imminent occurrence of the subsequent vehicle fault so that proactive maintenance is performed on the vehicle system.

Abstract: This factor analysis device is provided with a feature extraction unit (1021) that extracts feature quantities from an explanatory time series, a feature conversion unit (1022) that converts said feature quantities to a feature time series, a feature-time-series influence-degree computation unit (1031) that uses said feature time series and a response time series to compute an influence degree indicating the degree to which the feature time series influences the change over time represented by the response time series, and an explanatory-time-series influence-degree computation unit (1032) that uses said influence degree to compute an influence degree indicating the degree to which the explanatory time series influences the change over time represented by the response time series.

Abstract: A computer-implemented method for designing a built-in test is described. The method includes receiving, via a processor, a subsystem model including system parameters for a heat exchanger, wherein each of the system parameters includes a sensor variance; determining, via the processor, a test design vector based on one or more of the system parameters; and designing, via the processor, the built-in test based on the test design vector.

Abstract: Methods and systems for assessing, detecting, and responding to malfunctions involving components of autonomous vehicles and/or smart homes are described herein. Autonomous operation features and related components can be assessed using direct or indirect data regarding operation. Vehicle collision and/or smart home incident monitoring, damage detection, and responses are also described, with particular focus on the particular challenges associated with incident response for unoccupied vehicles and/or smart homes. Operating data associated with the autonomous vehicle and/or smart home may be received. Within the operating, an unusual condition indicative of a likelihood of incident may be detected. Based on the unusual condition, it may be determined that the incident occurred. Accordingly, a response to the incident may be determined. The response may be implemented by the autonomous vehicle and/or smart home.

Abstract: Negative path testing in a bootloader environment can include backing up a global state of a component under test, injecting a fault to trigger an error in the component under test in a bootloader environment, executing error handling instructions until a checkpoint of the component under test in the bootloader environment is reached, restoring the global state to the component under test from the backup, and restarting the component under test.

Abstract: A method for analyzing dysfunctions of an embedded system including a phase for modeling this system and an analysis phase, the modeling phase including defining a set of resources, a set of services, a set of statuses and at least one rule for the acquisition of a status by a resource or by a service, the analysis phase including configuring the analysis including marking a marked event and analyzing acquisition rules to determine a set of basic events and conditions for the appearance thereof, leading to the appearance of the marked basic event.

Abstract: Systems and methods providing actions and communications responsive to real-time events incorporating local, remote and learned information are disclosed. The system includes a cloud application on a cloud server and a premises device at a premises. The premises device has a plurality of sensors coupled to and/or included in the premises device. The cloud application receives signals, status and other information from the premises device. The cloud application also obtains information from third party information sources. The cloud application obtains location and other pertinent information about key persons. The cloud application evaluates actions to take in response to signals, status and information received from the premises device taking into consideration information from the third party information sources and information about key persons.

Abstract: A method for fault level estimation for a power converter quickly after a fault clearing event includes a step of providing a training dataset with an input variable set and a corresponding system fault level output variable. The method further includes a step of deducing a hypothesis function based on the training dataset and estimating the system fault level quickly after the fault clearing event using the hypothesis function. The hypothesis function is deduced either using a neural network computational model or using a multivariate polynomial regression approach. The estimated fault level can then be used to adjust the controller parameters of the power converter.

Abstract: A recording means records a value of pressure of fluid on a downstream side of a differential pressure generation means during a time period after a flow control valve open and control of a flow rate is started or after the flow control valve is closed and the control of a flow rate is stopped until a measured flow rate becomes stable, and a diagnostic means contrasts a controlled value acquired based in a value of the pressure of the fluid on the downstream side recorded by the recording means, with a controlled value when abnormality has not occurred, and diagnoses that abnormality has occurred in the flow meter when a difference between the both controlled values exceeds a threshold. Thereby, a mass flow controller and a diagnostic method which can simply diagnose generation of abnormality in a differential pressure type flow meter without requiring time only for performing the diagnosis.

Abstract: In an example embodiment, for each of a plurality of different entities in a social network structure, the entity is mapped into a vector having n coordinates. The vector for each of the plurality of different entities is stored in a deep representation data structure. One or more objective functions are applied to at least one combination of two or more of the vectors in the deep representation data structure. Then, an optimization test on each of the at least one combination of two or more of the vectors is performed using a corresponding objective function output for each of the at least one combination of two or more of the vectors, and, for any combination that did not pass the optimization test, one or more coordinates for the vectors in the combination are altered so that the vectors in the combination become closer together within an n-dimensional space.