Abstract: Implementations are described herein for leveraging digital media files retrieved and/or created by users to predict/determine topics of potential relevance to the users. In various implementations, digital media file(s) created and/or retrieved by a user with a client device may be applied as input across trained machine learning model(s), which in some cases are local to the client device, to generate output that indicates object(s) detected in the digital media file(s). Data indicative of the indicated object(s) may be provided to a remote computing system without providing the digital media file(s) themselves. In some implementations, information associated with the indicated object(s) may be retrieved and proactively output to the user. In some implementations, a frequency at which objects occur across a corpus of digital media files may be considered when determining a likelihood that a detected object is potentially relevant to a user.

Abstract: Conventional hierarchical time-series clustering is highly time consuming process as time-series are characteristically lengthy. Moreover, finding right similarity measure providing best possible hierarchical cluster is critical to derive accurate inferences from the hierarchical clusters. Method and system for Auto Encoded Compact Sequences (AECS) based hierarchical time-series clustering that enables compact latent representation of time-series using an undercomplete multilayered Seq2Seq LSTM auto encoder followed by generating of HCs using multiple similarity measures is disclosed. Further, provided is a mechanism to select the best HC among the multiple HCs on-the-fly, based on an internal clustering performance measure of Modified Hubert statistic ?. Thus, the method provides time efficient and low computational cost approach for hierarchical clustering for both on univariate and multivariate time-series.

Abstract: Media unit retrieval methods, systems and computer program products are provided that allow a user to search for an item by iteratively presenting media units such as images representing items to the user and receiving user input consisting of selections of the presented media units (including possibly the empty selection). Features, or attributes, a user is interested in, for example semantic features, are inferred from the interaction and media units are retrieved for presentation based on similarity with user-selected media units, through sampling of a probability distribution describing the intent or interests, or combinations of approaches. Accordingly, the user-experience is akin to a conversation about what the user is looking for. Retrieval may be based on both selected and unselected media units and the selection may comprise making a selection with a single action. Further, a database of media units can capture similarity relationships for efficient media unit retrieval.

Abstract: Embodiments of the present disclosure provide a method and apparatus for outputting information. A specific embodiment of the method includes: in response to receiving a query, detecting whether there is an entity slot in the query; in response to there being an entity slot in the query, adding the detected entity slot to a candidate slot; detecting, in the query, a relationship-determinative word of an entity; searching in a preset knowledge graph for a peripheral knowledge graph of the candidate slot; and inferring on the basis of the peripheral knowledge graph according to the relationship-determinative word, and outputting an entity word matching the relationship-determinative word.

Abstract: Systems and methods are provided for determining in real-time geographic areas having a threshold level of content consumption and deploying dynamic geo-fences to contain these geographic areas. These dynamic geo-fences provide a means for timing message notifications in order to optimize the chances of delivering targeted content to a mobile device user based on the current geographic location of the user's device relative to a threshold level of content consumption area. As mobile device users may be more likely to launch a client application in a place where other users are currently consuming content, a general message notification sent to the user's device located in a dynamic geo-fence created based on real-time content consumption, may increase the likelihood that the user will launch the client application and thereby, allow targeted content to be delivered to the user's mobile device.

Abstract: Disclosed are database systems, methods, systems, and computer program products for providing tabular data in a social network feed. In some implementations, a server of a database system stores, in a database, social network feed data comprising a plurality of feed items as data objects. The server also stores, in a database, tabular data as data objects. The server shares the social network feed data and the tabular data in a social network feed displayable to present the feed items and the tabular data in a user interface, the tabular data being presentable in the user interface in a tabular format. The user interface may receive shareable user commentary regarding the tabular data. The stored tabular data may be editable by users of the database system. The social network feed may be updated to comprise notifications of edits to the tabular data.

Abstract: A query specifying a source repository and a target repository is received from a client device. A source index is generated that corresponds to the source repository and represents a snapshot of metadata associated with data contained in the source data repository. The source index is filtered based on filtering criteria specified by the query to obtain a filtered source index. Attributes of data corresponding to the filtered source index are determined as well as data retrieval type parameters. Without initiating a data migration of the data corresponding to the filtered source index from the source repository to the target repository, predicted data migration metrics associated with the data migration are determined and presented to an end user of the client device. The end user is provided with the capability to initiate or forego the data migration based on an evaluation of the predicted data migration metrics.

Abstract: Systems and methods ingest machine data including logs, metadata, and cost and usage information from multiple heterogeneous cloud services. The machine data is saved as events. An application retrieves the metadata, events, metrics, and logs and causes an easy to understand visual representation of costs, resource usage, and non-compliance for each of a client's cloud services. Further, the data across the client's multiple heterogeneous cloud services is normalized to provide visual representations that compare the costs, resource usage, and non-compliance across the client's multiple heterogeneous cloud services. Further, machine learning aspects of the application can provide recommendations and trend analysis for cloud service asset usage.

Abstract: A computer system, computer program product, and a computer-implemented method for supplementing a data governance framework with one or more new data governance technical rules is disclosed. The method comprises providing a plurality of expressions and a first mapping. The expressions assign natural language patterns to technical language patterns. The first mapping maps first terms to data sources. A rule generator receives a new natural language (NL) rule comprising one or more natural-language patterns and one or more first terms. The rule generator resolves the new NL rule into one or more new technical rules interpretable by a respective rule engine and stores the one or more technical rules in a rule repository.

Abstract: Computer implemented reconstruction of compliance mapping due to an update in a regulation in the compliance mapping by a computing device includes comparing a first version of a regulation in the compliance mapping to a second, updated version of the first regulation. A change in the second version with respect to the first version is identified. The change may be an added control description, a deleted control description, or an updated control description. Upon determining that the change is an updated control description, the updated control description is analyzed to determine a type of update. The mapping of the regulation is reconstructed based on the change and, if the change is an updated control description, the type of update, using at least one of natural language processing and/or machine learning. The risk of the reconstructed mapping is assessed, and a service owner is notified about the risk of the changes.

Abstract: Disclosed is a system method for aligning a plurality of electronic map data, which includes the steps of obtaining an electronic map file, defining an area of interest on the electronic map file, selecting a predetermined datum and providing the electronic map file in the predetermined datum. Also disclosed is a land yield method including the steps of obtaining an electronic map, storing the electronic map in a computer readable medium, defining an area of interest on the electronic map, referring to a database to determine whether the database contains at least one restriction value, referring to a remote database to determine whether the remote database contains an additional restriction value, the additional restriction value being distinct from the at least one restriction value, combining restriction values, and determining a total value of the parcel of land based on the combined restriction values.

Abstract: An information retrieval system is provided.

Abstract: A method implements optimization of database queries by computing domain cardinality estimates. A client sends a database query to a server. The method parses the query to identify data columns. For each of the data columns, the method computes a lower bound and an upper bound of distinct data values using a pre-computed table size. The method also computes a patch factor by applying a pre-computed function to a ratio between a number of distinct data values that appear exactly once in a data sample and a number of distinct data values in the sample. Based on the patch factor, the lower bound, and the upper bound, the method computes an estimate of distinct values for each of the data columns. The method subsequently generates an execution plan for the query according to the computed estimates, executes the execution plan, and returns a result set to the client.

Abstract: A system executes a logo detection data model that processes, using computer processing resources, content data to detect a plurality of logos in the content data. The logo detection data model further automatically labels each detected logo of the plurality of logos in the content data.

Abstract: A system may include multiple client devices and a processing device communicatively coupled to the client devices. One or more client devices may implement a greedy approach in searching for an optimal artificial intelligence (AI) model. For example, a client device may use a training dataset to perform an AI task, and update its AI model. The client device may verify the performance of the AI task and determine whether to accept or reject its updated AI model. Upon rejection, the client device may repeat updating its AI model until the updated AI model is accepted, or until a stopping criteria is met. The processing device may be configured to update the initial AI models based on the accepted updated AI models obtained in the multiple client device. Training data for each of the client devices may contain a subset shuffled from a larger training dataset.

Abstract: A method, system, and computer program product, including generating a contrastive explanation for a decision of a classifier trained on structured data, highlighting an important feature that justifies the decision, and determining a minimal set of new values for features that alter the decision.

Abstract: One embodiment provides a method, including: receiving a multi-variate time-series dataset comprising a plurality of time-dependent datasets; for each of the plurality of time-dependent datasets, segmenting each of the plurality of time-dependent datasets at a transition point; clustering segments of the plurality of time-dependent datasets into clusters having similar lengths of segments; for each cluster (i) selecting a representative segment length and (ii) identifying a feature subset in that cluster; identifying, across the feature subsets, subset transition points, wherein each of the subset transition points corresponds to a change in value that meets a predetermined threshold within its corresponding feature subset; and determining, by applying a threshold test to the subset transition points, a segment length to be used in segmenting the entire multi-variate time-series dataset.

Abstract: An apparatus comprises a processing device configured to receive a query to extract information from a document, and to perform two or more iterations of utilizing a machine learning-based information extraction model to extract portions of unstructured text data from the document. In each iteration, a portion of the unstructured text data extracted from the document and an associated relevance score are output. In a first iteration, the query and document are input while in subsequent iterations the query and modified versions of the document are input, the modified versions having previously-extracted portions of the unstructured text data removed therefrom. The processing device is also configured to generate a response to the query comprising a subset of the portions of the unstructured text data extracted from the document determined to have associated relevance scores exceeding a threshold relevance score and at least a threshold level of similarity to the query.

Abstract: Certain aspects of the present disclosure provide techniques for operation of a feature management platform. A feature management platform is an end-to-end platform developed to manage the full lifecycle of data features. For example, to create a stateful feature, the feature management platform can receive a processing artifact from a computing device. The processing artifact defines the stateful feature, including the data source to retrieve event data from, when to retrieve the event data, the type of transform to apply, etc. Based on the processing artifact, the feature management system generates a processing job (e.g., the API defines a pipeline), which when initiated generates a vector that encapsulates the stateful feature. The vector is transmitted to the computing device that locally hosts a model, which generates a prediction that is transmitted to the feature management platform. Subsequently, the predication and stateful feature can be transmitted to other computing devices.

Abstract: An information handling system is provided for extracting data associated with a data field of a database from a randomly formatted source document. A processor of the information handling system receives the document, embeds a text-based representation of text of the document into vector data associated with the text, and encodes the vector data through a first neural network into reconstructed text activations. The processor further concatenates the vector data with an image-based representation of the document to provide first concatenated data, and encodes the first concatenated data through a second neural network into extracted visual feature activations. The processor further concatenates the reconstructed text activations with the extracted visual feature activations to provide second concatenated data, decodes the second concatenated data to identify particular data associated with the data field, and updates an entry of the database with the particular data in the data field.