Abstract: Various embodiments of the present disclosure include systems, methods, and non-transitory computer storage media configured to identify a set of training content items, each of the set of training content items comprising video content. A category may be assigned to each of the set of training content items. A plurality of variations may be provided to the each of the set of training content items. A first content recognition module may be trained in an unsupervised process to associate the plurality of variations of the each of the set of training content items with the category assigned to the each of the set of training content items. A classification layer may be generated based on the training the first content recognition module in the unsupervised process. A second content recognition module may be trained in a supervised process based on the classification layer.

Abstract: A computing system initializes a first frontier to be a root of a multi-dimensional hierarchical data structure representing an entity. The system acquires first data corresponding to the first frontier. The system performs modeling on the first data to obtain a first model and a corresponding first statistic. The system expands a dimension of the first frontier. The system gathers second data corresponding to the expanded frontier. The system applies the data modeling on the second data to obtain a second model and a corresponding second statistic. The system compares the first statistic of the first model and the second statistic of the second model. The system sets the second model to be the first model in response to determining that the second model statistic is better than the first model statistic. The system outputs the first model.

Abstract: Embodiments of the present invention relate to meeting latency constraints in a multi-core neurosynaptic network. In one embodiment of the present invention, a method of and computer program product for power-driven synthesis under latency constraints is provided. Power consumption of a neurosynaptic network is modeled as wire length. The neurosynaptic network comprises a plurality of neurosynaptic cores. Each of the plurality of neurosynaptic cores is modeled as a node in a placement graph. The graph has a plurality of edges. A weight is assigned to each of the plurality of edges based on a spike frequency. An arrangement of the neurosynaptic cores is determined. The arrangement comprises a length of each of the plurality of edges. A maximum length is compared to the length of each of the plurality of edges. The weight of at least one of the plurality of edges is increased where the length is greater than the maximum length.

Abstract: Users of social networking system are provided with user interface elements permitting the user to post comments on pages within the social networking system. Pages may be provided for any non-user entity, including for example, pages for businesses, products, concepts, etc. Embodiments provided herein permit page moderators to ban certain content from being displayed on a page. For example, the social networking system may receive a list of proscribed content and block comments containing the proscribed content by reducing the distribution of those comments to other viewing users. However, the social networking system may display the blocked content to the commenting user such that the commenting user is not made aware that his or her comment was blocked, thereby providing fewer incentives to the commenting user to spam the page or attempt to circumvent the social networking system filters.

Abstract: Methods and systems for multi-tiered information retrieval training are disclosed. A method includes identifying results in a ranked ordering of results that can be swapped without changing a score determined using a first ranking quality measure, determining a first vector and at least one other vector for each identified swappable result in the ranked ordering of results based on the first ranking quality measure and at least one other ranking quality measure respectively, and adding the first vector and the at least one other vector for each identified swappable result in the ranked ordering of results to obtain a function of the first vector and the at least one other vector. Access is provided to the function of the first vector and the at least one other vector for use in the multi-tiered information retrieval training.

Abstract: System and techniques for temporally encoded static spatial images are described herein. A static spatial image may be obtained. Here, the static spatial image defines pixel values over an area. A scan path may be selected. Here, the scan path defines a path across the area of the static spatial image. A window is scanned (e.g., moved or slid) along the scan path on the static spatial image to produce changes in a portion of the window over time. The changes in the portion of the window are recorded along with respective times of the changes.

Abstract: A recurrent neural network (RNN) is trained to identify split positions in long content, wherein each split position is a position at which the theme of the long content changes. Each sentence in the long content is converted to a vector that corresponds to the meaning of the sentence. The sentence vectors are used as inputs to the RNN. The high-probability split points determined by the RNN may be combined with contextual cues to determine the actual split point to use. The split points are used to generate thematic segments of the long content. The multiple thematic segments may be presented to a user along with a topic label for each thematic segment. Each topic label may be generated based on the words contained in the corresponding thematic segment.

Abstract: A medical scan image analysis system is operable to receive a plurality of medical scans that represent a three-dimensional anatomical region and include a plurality of cross-sectional image slices. A plurality of three-dimensional subregions corresponding to each of the plurality of medical scans are generated by selecting a proper subset of the plurality of cross-sectional image slices from each medical scan, and by further selecting a two-dimensional subregion from each proper subset of cross-sectional image slices. A learning algorithm is performed on the plurality of three-dimensional subregions to generate a fully convolutional neural network. Inference data corresponding to a new medical scan received via the network is generated by performing an inference algorithm on the new medical scan by utilizing the fully convolutional neural network. An inferred abnormality is identified in the new medical scan based on the inference data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for optimizing digital component transmission. A data structure stores session data for a user session. An encoder accesses the session data, encodes at least a portion of the online activities by representing different online activities with different symbols, and generates an encoded session string that includes multiple different symbols representing an order of occurrence of the different online activities. One or more servers classify the user session to a positive session classification or a negative session classification based on a classification score for the encoded session string, aggregates the classification of the user session with other classifications of other user sessions that have a same session string as the encoded session string, and adjusts distribution of digital components to client devices based on the aggregated classifications.

Abstract: A method for detecting a data flow type includes obtaining a header of a first data packet of a current data flow and a pattern vector of the current data flow from the header; comparing the at least one feature dimension in the pattern vector of the current data flow with a corresponding feature dimension in a pattern vector of at least one historical data flow, so as to obtain at least one pattern similarity of the current data flow; predicting a length of the current data flow according to the at least one pattern similarity of the current data flow and a length of the corresponding at least one historical data flow; and comparing the predicted length of the current data flow with a preset threshold, and determining whether the current data flow is a large data flow or a small data flow according to a comparison result.

Abstract: Examples herein relate to hardware accelerators for calculating node values of neural networks. An example hardware accelerator may include a crossbar array programmed to calculate node values of a neural network and a current comparator to compare an output current from the crossbar array to a threshold current according to an update rule to generate new node values. The crossbar array has a plurality of row lines, a plurality of column lines, and a memory cell coupled between each unique combination of one row line and one column line, where the memory cells are programmed according to a weight matrix. The plurality of row lines are to receive an input vector of node values, and the plurality of column lines are to deliver an output current to be compared with the threshold current.

Abstract: An approach is described for adjusting prominence of a participant profile in a social networking interface. An associated system may include a processor and a memory storing an application program, which, when executed on the processor, performs an operation that may include receiving an activity stream update of the participant and calculating a relevancy score based on content in the activity stream update. The operation further may include adjusting a visibility level of the participant profile in the social networking interface based upon the calculated relevancy score. Adjusting the visibility level may include increasing the visibility level of the participant profile upon determining that the calculated relevancy score is greater than or equal to a first predefined threshold value. Adjusting the visibility level further may include decreasing the visibility level of the participant profile upon determining that the calculated relevancy score is less than a second predefined threshold value.

Abstract: A method for pooling at least one ROI by using one or more masking parameters is provided. The method is applicable to mobile devices, compact networks, and the like via hardware optimization. The method includes steps of: (a) a computing device, if an input image is acquired, instructing a convolutional layer of a CNN to generate a feature map corresponding to the input image; (b) the computing device instructing an RPN of the CNN to determine the ROI corresponding to at least one object included in the input image by using the feature map; (c) the computing device instructing an ROI pooling layer of the CNN to apply each of pooling operations correspondingly to each of sub-regions in the ROI by referring to each of the masking parameters corresponding to each of the pooling operations, to thereby generate a masked pooled feature map.

Abstract: Methods, computer-readable media, and systems are disclosed for applying 1D processing in a non-1D formation. In some embodiments, a 3D model or curtain section of a subsurface earth formation may be obtained. A processing window within the 3D model or curtain that is suitable for 1D inversion processing is determined, and a local 1D model for the processing window is built. A 1D inversion is performed on the local 1D model, and inverted formation parameters are used to update the 3D model or curtain section.

Abstract: Methods and apparatus for optimizing computer detection of malware using pattern recognition by refreshing random classification forests are described. In one embodiment, the method may include building a random forest with two or more binary decision trees based at least in part on a first set of categorized data, sending the random forest to a client device with a first random forest control value, identifying a second set of categorized data different from the first set of categorized data, calculating a second random forest control value based on the second set of categorized data and sending the second random forest control value to the client device.

Abstract: An approach is provided in which a knowledge manager trains a machine-learning model and generates a hyperplane based upon a first set of labeled feature vectors. The knowledge manager computes, relative to the hyperplane, a first distribution of a first set of feature vectors corresponding to a first set of source documents. Subsequently, the knowledge manager computes, relative to the hyperplane, a second distribution of a second set of feature vectors corresponding to a second group of source documents. The knowledge manager, in turn, generates an indicator to retrain the machine-learning model in response to determining that a distribution difference between the second distribution and the first distribution reaches a distribution difference threshold.

Abstract: A learning method for segmenting an image having one or more lanes is provided to be used for supporting collaboration with HD maps required to satisfy level 4 of autonomous vehicles. The learning method includes steps of: a learning device instructing a CNN module (a) to apply convolution operations to the image, thereby generating a feature map, and apply deconvolution operations thereto, thereby generating segmentation scores of each of pixels on the image; (b) to apply Softmax operations to the segmentation scores, thereby generating Softmax scores; and (c) to (I) apply multinomial logistic loss operations and pixel embedding operations to the Softmax scores, thereby generating Softmax losses and embedding losses, where the embedding losses is used to increase inter-lane differences among averages of the segmentation scores and decrease intra-lane variances among the segmentation scores, in learning parameters of the CNN module, and (II) backpropagate the Softmax and the embedding losses.

Abstract: Systems and methods are provided for analyzing images or video using computer vision. Media items, including images or videos, are retrieved from one or more media channels, such as social media network services, streaming media networks, or broadcast networks. A media item is identified that depicts a object associated with a sponsor. A media cost equivalent is determined that is associated with the media item as distributed by the media channels. Using computer vision, a percentage is determined for the media cost equivalent. The percentage is determined based at least in part on at least one of size, clarity, duration or position of the sponsor object or another object in the media item. A sponsor valuation is calculated based at least in part on the media cost equivalent adjusted by the percentage.

Abstract: A method of facilitating simulations of industrial processes is disclosed. The method can be applied to the simulation of hydrocarbon processing, including oil and gas processing and production, refining and petrochemicals processing. The method includes receiving process information defining a process for simulation; creating and storing at least one rule defining a time-dependent property of the process information; and simulating the process based on the received process information under variation of the time-dependent property of the process information. An associated apparatus is also disclosed.

Abstract: Systems and methods are provided for identifying a set of highly-correlated genes for use in classifying both a first type of biological sample and a second type of biological sample as to a phenotypic characterization, where the first type of biological sample and the second type of biological sample are a sample type selected from the group consisting of a fresh biological sample, a frozen biological sample, a biological sample that has been preserved with a non-crosslinking preservative, and a fixed biological sample that has been fixed with a crosslinking agent, and where the first type of biological sample and the second type of biological sample are sample types that are not the same sample type. The invention also relates to computer systems and methods for use in training a classifier using the highly-correlated genes, and using the trained classifier to classify biological samples that are of the same sample type as the first type of biological sample or the second type of biological sample.

Abstract: Sustainable building lighting and energy modelling and control, and the associated computer graphics, including real-time dynamic lighting simulation, are concerned with: an optimized method for radiance modelling, including its application to predictive daylight harvesting; and the real-time simulation of physically-based electric lighting and daylighting for architectural, horticultural, and theatrical lighting systems visualization. In order to display and analyze in real time a photometrically accurate representation of an environment, thousands of lighting channels may have their intensity settings continually varied such that a user may interactively view the three-dimensional environment without the need for ongoing global illumination calculations. This can be accomplished utilizing texture maps as a multiplicity of canonical radiosity solutions, each representing a lighting channel for dynamic lighting simulation, and storing the solutions in the texture memory of a graphics processing unit.

Abstract: Methods and apparatus for optimizing computer detection of malware using pattern recognition by refreshing random classification forests are described. In one embodiment, one or more selected trees of a random forest on a computing system may be replaced by one or more new trees. As new categorized data becomes available, one or more new trees may be generated using the new categorized data. Once the one or more new trees are available, the performance of the one or more new trees may be compared to the performance of the trees in the current random forest. Based on this comparison, one or more trees of the random forest may be selected to be replaced by one or more of the new trees.

Abstract: A method and system are provided. The method includes deriving a set of user attributes from an aggregate analysis of images and videos of a user. The deriving step includes recognizing, by a set of visual classifiers, semantic concepts in the images and videos of the user to generate visual classifier scores. The deriving step further includes deriving, by a statistical aggregator, the set of user attributes. The set of user attributes are derived by mapping the visual classifier scores to a taxonomy of semantic categories to be recognized in visual content. The deriving step also includes displaying, by an interactive user interface having a display, attribute profiles for the attributes and comparisons of the attribute profiles.

Abstract: The present invention relates to a method for incremental learning of a classification model, where pre-defined weak incremental learners are distributed over the distinct regions in a set of partitionings of the input domain. The partitionings and regions are organized via a binary tree and they are allowed to vary in a data-driven way, i.e., in a way to minimize the classification error rate. Moreover, to test a given data point, a mixture of decisions is obtained through the models learned in the regions that this point falls in. Hence, naturally, in the cold start phase of the data stream, the simpler models belonging to the larger regions are favored and as more data get available, the invention automatically puts more weights on the more complex models.

Abstract: A system, method, and apparatus are provided for depicting attributes of a user's or member's connections within a professional or social network, in a manner that enables insight into the nature of the connections. When insight into a first member's connections is to be depicted, one or more attributes of first-degree connections (i.e., direct connections) and/or other connections are retrieved (e.g., employer or associated organization, function or role, seniority, skill(s)), over all time or some threshold period of time into the past. The values for the attributes are aggregated over the examined connections, and some or all values are displayed with one or more visual characteristics (e.g., size, color) that indicate a frequency or commonness of the values among the connections. The system or apparatus may include a user interface through which the depiction is presented in place of or in addition to a display of the connections.

Abstract: Embodiments are disclosed for determining whether a computing node is in a normal or an abnormal condition based on its characteristics relative to those of other computing nodes. In embodiments, log files for the computing node are used to develop a state model of the computing node, and where the state model differs between two similar computing nodes, an abnormality is identified. In other embodiments, characteristics about computing nodes (e.g., CPU resources used) are used to cluster those computing nodes, and those computing nodes that lie outside of a cluster are identified as abnormal.

Abstract: According to an example, configurable network security may include receiving data flows directed to end node modules of a server, and selecting data flows from the received data flows based on an analysis of attributes of the received data flows. The selected data flows may be less than the received data flows. A number of IPS data plane modules of the server that are available for inspection of the selected data flows may be determined. The selected data flows may be distributed between the IPS data plane modules based on the determined number of the IPS data plane modules. The distributed data flows may be inspected using the IPS data plane modules to identify malicious and benign data flows, and to determine whether to drop the malicious data flows, direct the malicious data flows to a predetermined destination, or forward the benign data flows to the end node modules.

Abstract: A system for verifying resource propagations in real-time typically includes a classical computer apparatus and a quantum optimizer in communication with the classical computer apparatus. The quantum optimizer is configured to analyze resource propagation information related to previous resource propagations and confirmed attacks to generate a model for verifying resource propagations or mitigating potential attacks. When the classical computer apparatus receives indication of a present resource propagation, the classical computer apparatus transfers attributes of the present resource propagation to the quantum optimizer, which analyzes the resource propagation using the model to determine whether the resource propagation involves a potential attack. Based on the indication from the quantum optimizer, the classical computer apparatus attempts to mitigate the potential attack or allows the resource propagation to proceed.

Abstract: An information processing device includes: an outline extraction unit extracting an outline of a subject from a picked-up image of the subject; a characteristic amount extraction unit extracting a characteristic amount, by extracting sample points from points making up the outline, for each of the sample points; an estimation unit estimating a posture of a high degree of matching as a posture of the subject by calculating a degree of the characteristic amount extracted in the characteristic amount extraction unit being matched with each of a plurality of characteristic amounts that are prepared in advance and represent predetermined postures different from each other; and a determination unit determining accuracy of estimation by the estimation unit using a matching cost when the estimation unit carries out the estimation.

Abstract: A computer system and computer implemented method for extracting data set from data clusters that comprises of rows and columns of heterogeneous data values. A plurality of random data groups comprising of at least one of a plurality of contiguous row or columns of data values are selected. Each data value has a data type. A table template type is identified based on detection of a pattern between the data cells of the contiguous rows or columns. A table template header is identified that comprises of a starting position, and ending position and a width. A reference row or reference column indicating a start of a table body is determined. The data cells of the subsequent rows or columns in the table body are compared to the data cells of the reference rows to identify noise rows or columns that are removed from the table body.

Abstract: Content that is significant to a user may be determined. An indication that a user finds content within a document significant may be received. In response to the received indication, the document may be analyzed to identify a set of topics associated with the content of the document. From the set of topics, a subset of topics responsible for the user finding the document significant may be identified. A user profile associated with the user may be updated based on the subset of topics.

Abstract: Various embodiments of the present technology include methods of assessing risk of a cyber security failure in a computer network of an entity. Some embodiments involve using continual or periodic data collecting to improve inferential analysis, as well as obtaining circumstantial or inferential information from social networks. Machine learning may be used to improve predicitive capabilities. Some embodiments allow for identification of an entity from circumstantial or inferential information based on the machine learning and comparative analyses.

Abstract: Group aware command based arrangement is provided. An alignment command can be received, an undirected graph comprising a group of graph nodes connected by graph edges can be constructed, wherein the group of graph nodes represent a group of selected display elements, and the undirected graph can be partitioned as a function of identifying a severable graph edge of the graph edges that connects a pair of graph nodes included in the group of graph nodes to form a first collection of graph nodes and a second collection of graph nodes.

Abstract: A current-mode PUF circuit based on a reference current source comprises an input register, the reference current source, a deviation current comparator and a timing controller. The input register is used for ensuring synchronization of the input challenges to avoid influences of asynchronous challenges on output responses. The reference current source generates a reference current for temperature and voltage compensation. A deviation current source array generates two paths of deviation currents under the control of the input challenges. The deviation current comparator generates and outputs a judgement according to the magnitude of the current provided by the deviation current source array. The timing controller is used for generating timing information for operation of the PUF circuit. The invention has the following advantages: the deviation current source array outputs deviation currents with high robustness and high reliability.

Abstract: Dynamically developing and maintaining threat models, threat landscapes and threat matrices are described. Specifically described are techniques on how to relate: (1) attack surfaces, (2) attack histories, (3) threats and (4) historical responses, by loading these four types of data, as well as other data, into a data store. One example data store disclosed includes some variations of a graph data structure. Upon loading the data, the populated data store may be used to develop Threat Models that will represent a Threat Landscape and a Threat Matrix. These may then be queried for recommended reactive and proactive responses with respect to an installation, in order to improve security.

Abstract: Optimizing temporal pulse patterns for stimulation delivery to a subject includes: generating a first generation pulse pattern; delivering stimulation according to the first generation pattern to a subject; measuring efficacy, efficiency and side-effect parameters affected by the delivered stimulation; determining a fitness of the first generation pattern using the measured parameters; generating a second generation pattern using the first generation pattern according to the determined fitness of the first generation temporal pattern; and delivering stimulation according to the second generation pattern. Iterative further optimization may include crossing any particular generation temporal pulse pattern with at least one other temporal pulse pattern to generate offspring patterns for further use and optimization. Immigrant random patterns may be added, and offspring patterns may receive point mutations.

Abstract: Techniques for mapping item listings from a first taxonomy to a second taxonomy are described. In an example embodiment, item listings from a first database storing a first taxonomy and item listings from a second database storing a second taxonomy are obtained. Then, for each of the obtained item listings, a plurality of features is extracted, including at least one feature related to an image associated with the item listing and at least one feature related to text associated with the item listing. Then a mapping between item listings in the first taxonomy and item listings in the second taxonomy is created based on the plurality of features extracted by the feature extraction component, wherein the mapping identifies which item listings in the first taxonomy correlate to a same product as which item listings in the second taxonomy.

Abstract: The disclosed computer-implemented method for determining whether malicious files are targeted may include (i) applying, to a malware detection structure, a plurality of sample data points, each sample data point corresponding to at least one of a malicious file known to be targeted and a malicious file known to be non-targeted, (ii) identifying one or more boundaries of the sample data points within the malware detection structure, (iii) determining, after identifying the sample boundaries, that a new data point falls outside of the boundaries, and (iv) classifying a malicious file associated with the new data point as non-targeted in response to determining that the new data point falls outside of the sample boundaries. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: Embodiments include identifying, at a logical path node, a first logical path and a second logical path; executing, by a processor implemented at least partially in hardware, a first set of instructions to follow the first logical path; storing, in a memory, a first set of information obtained from following the first logical path; evaluating, by a malware handler module implemented at least partially in hardware, the first set of information for malware; restoring, from the memory, environmental data for the first logical path node; executing, by the processor, a second set of instructions to follow the second logical path; storing, in a memory, a second set of information obtained from following the second logical path; and evaluating, by the malware handler module, the second set of information for malware.

Abstract: A configuration engine traverses sequential levels of a decision tree in order to iteratively refine a configuration for a solar power system. At each level of the decision tree, the configuration engine determines the outcome of a design decision based on computing the result of a value function. The configuration engine explores configurations that optimize the value function result compared to other configurations, and may also discard less optimal configurations. When a current configuration is considered less optimal than a previous configuration generated at a previous level, the configuration engine discards the current configuration and re-traverses the decision tree starting with the previous configuration.

Abstract: A method and apparatus are provided for positioning an unmanned robotic vehicle (URV). The URV captures a set of one or more of image and non-image information of an object while positioned at a first position, provides the set of image/non-image information to a server entity, in response to providing the set of image/non-image information, receives a three-dimensional (3D) model associated with the object, autonomously determines a second position based on the 3D model, and autonomously navigates to the second position. At the second position, the URV may capture further image and/or non-image information and, based on the further captured image/non-image information, autonomously determine, and navigate to, a third position. The steps of capturing further image and/or non-image information and, based on the captured image and/or non-image information, autonomously determining and navigating to further positions may be repeated indefinitely, or until otherwise instructed.

Abstract: A method for analyzing three-dimensional image data of a target region of a subject, the method comprising identifying a region of interest within the target region containing imaging information predictive of a disease state of the target region, calculating at least two radiomic features associated with the region of interest, and classifying the region of interest based on the calculated radiomic features.

Abstract: A method of identifying a spatial distribution of areas includes: a first step of determining a rough spatial distribution of at least one area in a sensory area of cerebral cortex in each of a left hemisphere and right hemisphere of the brain that is a target area of the identification; a second step of applying a sensory stimulus to the identification target area; a third step of obtaining brain activity information; and a fourth step of using the brain activity information to calculate one of a coefficient of cross-correlation, and coherence, between brain activity information of areas, assessing synchrony between the left and right hemispheres, and changing the spatial distribution determined in the first step.

Abstract: A governance apparatus and a communication method for communicating within the governance apparatus. The governance apparatus includes a Government. The Government includes Councils such that a macro grid including an artificial intelligence and the Government is configured to respond to an alert pertaining to an event through use of the artificial intelligence and the Government. The governance apparatus also includes an enhanced Transmission Control Protocol/Internet Protocol (TCP/IP) communication stack of layers including a Governance Layer and an Intelligence Layer. The Intelligence Layer includes intelligence software configured to process data pertaining to the event, data pertaining to the alert, and data pertaining to the Government.

Abstract: Historical telemetry data can be used to generate predictions for various classes of data at various aggregates of a system that implements an online service. An anomaly detection process can then be utilized to detect anomalies for a class of data at a selected aggregate. An example anomaly detection process includes receiving telemetry data originating from a plurality of client devices, selecting a class of data from the telemetry data, converting the class of data to a set of metrics, aggregating the set of metrics according to a component of interest to obtain values of aggregated metrics over time for the component of interest, determining a prediction error by comparing the values of the aggregated metrics to a prediction, detecting an anomaly based at least in part on the prediction error, and transmitting an alert message of the anomaly to a receiving entity.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for query analysis. Queries are identified in query data, and an entity-descriptive portion and a suffix are determined in each query. Query counts are determined for a number of times that the respective queries occur in the query data. Based on the query counts, an entity-level count is estimated, which represents a number of query submissions that include the particular suffix and are considered to refer to a first entity. The entity is determined to be a particular type of entity. A type-level count is determined, which represents a number of query submissions that include the first suffix and are estimated to refer to entities of the particular type of entity. A score is assigned to the particular suffix based on the entity-level count and the type-level count.

Abstract: A method and apparatus for locating dot text in an image are described. A set of dots is extracted. A determination of whether a first region of interest (ROI) including the set of dots satisfies selection criteria is performed, where the first region of interest is oriented based on results from a principal component analysis of the set of dots. Responsive to determining that the first ROI does not satisfy the selection criteria, performing the following: removing an outlier dot from the first set of dots to obtain a second set of dots; when the second ROI satisfies the selection criteria, outputting the second ROI as a location of the dot text in the image, and when the second region of interest does not satisfy the selection criteria, repeating the operations until a resulting ROI is determined to satisfy the selection criteria.

Abstract: Embodiments relate to systems, devices, and computer-implemented methods for characterizing domain names by determining a name server switching footprint for domain names using a data set corresponding to name server operations for the domain names. The domain names can be clustered into groups based on the name server switching footprints, and intended uses of domain names in a group can be extrapolated to other domain names in the group. Name server switching footprints can also be predicted for new domains names using a prediction model trained using the determined name server switching footprints for the domain names in the data set.

Abstract: A quality-directed adaptive analytic retraining is provided. Training example data with which to retrain a machine learning model that has been previously trained is received. The training example data is stored in a memory. The machine learning model is evaluated at least by running the machine learning model with the training example data. A normalized quality measure may be determined based on the evaluating. Whether to retrain the machine learning model is determined at least based on the normalized quality measure. Responsive to determining that the machine learning model is to be retrained, the machine learning model is retrained.

Abstract: In one embodiment, a method includes receiving usage data regarding usage of a subscription-based media service account, generating a plurality of sharing scores based on the usage data, each of the plurality of sharing scores being indicative of a confidence that the usage of the subscription-based media service account is subject to a respective type of sharing, comparing each of the plurality of sharing scores to a respective threshold, and, in response to one of the plurality of sharing scores exceeding its respective threshold, presenting a challenge associated with the respective type of sharing.