Abstract: Techniques are described for automatically and dynamically modifying ongoing operations of computing devices in device-specific manners, such as based on an automated identification of a computing device's status (e.g., identifying a likely ongoing or imminent failure of a smart phone or other computing device based on a series of observed hardware states of the computing device, and taking automated corrective actions to prevent or otherwise mitigate such device failure, such as by modifying configuration settings on the computing device or on associated systems). The techniques may include, for each of multiple device status outcomes of interest (e.g., device failure versus device non-failure), generating a state-space outcome model representing devices that reach that status outcome within a time period of interest, and using such outcome models to identify a likely ongoing or imminent outcome of a current device, with corresponding automated corrective actions then taken.

Abstract: Techniques are described for automatically and dynamically modifying ongoing operation of computing devices in device-specific manners, such as to improve ongoing performance of the computing devices by modifying configuration settings on the computing devices or on associated systems in communication with the computing devices. The techniques may include generating one or more decision structures that are each specific to a type of measured performance effect, and using the decision structure(s) to improve corresponding performance of a computing device, with the generating of the decision structure(s) including analyzing training data that associates prior measured performance effects with corresponding attributes of computing devices and of modification actions that were performed for the computing devices.

Abstract: Techniques are disclosed that automatically identify and order the most differentiated clusters from a given collection of clusters within a dataset. A measure of dissimilarity is computed for each cluster from a defined reference cluster, and the clusters are ordered according to the chosen dissimilarity. At least N clusters are selected as the most differentiated clusters relative to the defined reference. Within each cluster, the top-M most distinguishing cluster attributes can be automatically identified by an analogous process that computes the dissimilarity of each cluster attribute to its corresponding attribute in the reference cluster, and orders the attributes by dissimilarity. This then allows for automatic surfacing of what it is about a cluster that differentiates its members relative to the population as a whole, and to provide insight on what action or treatment might be made to address that specific segment of the underlying population.

Abstract: Techniques train a tree to identify offers to send to a particular customer. Messages that include offers and having attributes are sent to a target user group. Feature measure results from the messages on the target user group, is used with feature measure results for a control user group, to train the tree with branch splits being identified based on maximizing an information gain from the feature measure results for a message/user attribute, where each node within the tree includes target and control distributions for the feature measure. The tree is traversed for a given marketing message/user, drawing randomly from feature measure distributions in the tree to determine whether to send the given marketing message to the user. By drawing randomly from the feature measure distributions, exploration and exploitation of various messages may be performed to minimize ignoring of messages that may have an information gain for particular customers.

Abstract: Techniques are disclosed that automatically identify and order the most differentiated clusters from a given collection of clusters within a dataset. A measure of dissimilarity is computed for each cluster from a defined reference cluster, and the clusters are ordered according to the chosen dissimilarity. At least N clusters are selected as the most differentiated clusters relative to the defined reference. Within each cluster, the top-M most distinguishing cluster attributes can be automatically identified by an analogous process that computes the dissimilarity of each cluster attribute to its corresponding attribute in the reference cluster, and orders the attributes by dissimilarity. This then allows for automatic surfacing of what it is about a cluster that differentiates its members relative to the population as a whole, and to provide insight on what action or treatment might be made to address that specific segment of the underlying population.

Abstract: Techniques train a tree to identify offers to send to a particular customer. Messages that include offers and having attributes are sent to a target user group. Feature measure results from the messages on the target user group, is used with feature measure results for a control user group, to train the tree with branch splits being identified based on maximizing an information gain from the feature measure results for a message/user attribute, where each node within the tree includes target and control distributions for the feature measure. The tree is traversed for a given marketing message/user, drawing randomly from feature measure distributions in the tree to determine whether to send the given marketing message to the user. By drawing randomly from the feature measure distributions, exploration and exploitation of various messages may be performed to minimize ignoring of messages that may have an information gain for particular customers.

Abstract: Techniques disclosed herein employ entity-activity data expressed in a discrete distribution (histogram) form having one or many dimensions to dynamically classify the entity's usage and/or behavior patterns, where groupings or segmentations of different entities that exhibit similar usage patterns are identified using various approaches, including dimensionality reduction, and/or clustering procedures. A consensus or ensemble clustering may be generated that represents a clustering of clusters, based on subclusterings themselves, and/or any combination of subclusters with entity-activity data to selectively execute a market offering campaign. In one embodiment, the resulting ensemble clusterings enable selective directing of targeted offerings to a telecommunication provider's customers.

Abstract: Techniques disclosed herein employ entity-activity data expressed in a discrete distribution (histogram) form having one or many dimensions to dynamically classify the entity's usage and/or behavior patterns, where groupings or segmentations of different entities that exhibit similar usage patterns are identified using various approaches, including dimensionality reduction, and/or clustering procedures. A consensus or ensemble clustering may be generated that represents a clustering of clusters, based on subclusterings themselves, and/or any combination of subclusters with entity-activity data to selectively execute a market offering campaign. In one embodiment, the resulting ensemble clusterings enable selective directing of targeted offerings to a telecommunication provider's customers.

Abstract: Techniques are disclosed that automatically identify and order the most differentiated clusters from a given collection of clusters within a dataset. A measure of dissimilarity is computed for each cluster from a defined reference cluster, and the clusters are ordered according to the chosen dissimilarity. At least N clusters are selected as the most differentiated clusters relative to the defined reference. Within each cluster, the top-M most distinguishing cluster attributes can be automatically identified by an analogous process that computes the dissimilarity of each cluster attribute to its corresponding attribute in the reference cluster, and orders the attributes by dissimilarity. This then allows for automatic surfacing of what it is about a cluster that differentiates its members relative to the population as a whole, and to provide insight on what action or treatment might be made to address that specific segment of the underlying population.

Abstract: Techniques are disclosed that leverage time series techniques to express entity-activity data in a longitudinal temporal form, which may then be employed to dynamically classify the entity's behavior. In some embodiments, groupings or segmentations of different entities that exhibit similar profiles of longitudinal temporal form are identified using various techniques, including frequency-domain analysis, and/or unsupervised model-based clustering. The clustering of entities enables directing of offerings to, for example, a telecommunication's customer based on characteristics of the cluster.

Abstract: Copies of original sound recordings are identified by extracting features from the copy, creating a vector of those features, and comparing that vector against a database of vectors. Identification can be performed for copies of sound recordings that have been subjected to compression and other manipulation such that they are not exact replicas of the original. Computational efficiency permits many hundreds of queries to be serviced at the same time. The vectors may be less than 100 bytes, so that many millions of vectors can be stored on a portable device.

Abstract: Copies of original sound recordings are identified by extracting features from the copy, creating a vector of those features, and comparing that vector against a database of vectors. Identification can be performed for copies of sound recordings that have been subjected to compression and other manipulation such that they are not exact replicas of the original. Computational efficiency permits many hundreds of queries to be serviced at the same time. The vectors may be less than 100 bytes, so that many millions of vectors can be stored on a portable device.

Abstract: Copies of original sound recordings are identified by extracting features from the copy, creating a vector of those features, and comparing that vector against a database of vectors. Identification can be performed for copies of sound recordings that have been subjected to compression and other manipulation such that they are not exact replicas of the original. Computational efficiency permits many hundreds of queries to be serviced at the same time. The vectors may be less than 100 bytes, so that many millions of vectors can be stored on a portable device.

Abstract: Copies of original sound recordings are identified by extracting features from the copy, creating a vector of those features, and comparing that vector against a database of vectors. Identification can be performed for copies of sound recordings that have been subjected to compression and other manipulation such that they are not exact replicas of the original. Computational efficiency permits many hundreds of queries to be serviced at the same time. The vectors may be less than 100 bytes, so that many millions of vectors can be stored on a portable device.