Abstract: The present invention relates to methods for evaluating and/or predicting the outcome of a clinical condition, such as cancer, metastasis, AIDS, autism, Alzheimer's, and/or Parkinson's disorder. The methods can also be used to monitor and track changes in a patient's DNA and/or RNA during and following a clinical treatment regime. The methods may also be used to evaluate protein and/or metabolite levels that correlate with such clinical conditions. The methods are also of use to ascertain the probability outcome for a patient's particular prognosis.

Abstract: A computer-implemented technique is described herein for detecting actionable items in speech. In one manner of operation, the technique can include receiving utterance information that expresses at least one utterance made by one participant of a conversation to at least one other participant of the conversation. The technique can also include converting the utterance information into recognized speech information and using a machine-trained model to recognize at least one actionable item associated with the recognized speech information. The technique can also include performing at least one computer-implemented action associated the actionable item(s). The machine-trained model may correspond to a deep-structured convolutional neural network. The technique can produce the machine-trained model using a source environment corpus that is not optimally suited for a target environment in which the model is intended to be applied.

Abstract: In general, embodiments of the present invention provide systems, methods and computer readable media for automated dynamic data quality assessment. One aspect of the subject matter described in this specification includes the actions of receiving a data quality job including a new data sample; and, if the new data sample is determined to be added to a reservoir of data samples, sending a quality verification request to an oracle; receiving a new data sample quality estimate from the oracle; and adding the new data sample and estimate to the reservoir. A second aspect of the subject matter includes the actions of receiving, from a predictive model, a judgment associated with a new data sample; analyzing the new data sample based in part on the judgment to determine whether to send a new data sample quality verification request to an oracle; and, if a new data sample quality estimate is received from the oracle, determining whether to add the new data sample and the judgment to the reservoir.

Abstract: A data model is traversed to determine concept characteristics associated with concepts that may be associated with entities. Associated documents may be evaluated to identify document characteristics associated with the entities. Entity models may be trained based on the concept characteristics and the document characteristics with each entity model being associated with a confidence value. Results for one or more queries based on the documents and the entity models may be provided. The results may reference the documents that may be associated with the entities. Some entity models may produce results that have a confidence value below a threshold value. Accordingly, the entity models that provide low confidence results may be re-trained.

Abstract: The operation of an application on a first device may be guided by a user operating a second device. The application on the first device may present a character on a display of the first device and obtain an audio signal of speech of a user of the first device. Audio data may be transmitted to the second device and corresponding audio may be played from speakers of the second device. The second device may present suggestions of phrases to be spoken by the character displayed on the first device. A user of the second device may select a phrase to be spoken by the character. Phrase data may be transmitted to the first device, and the first device may generate audio of the character speaking the phrase using a text-to-speech voice associated with the character.

Abstract: At a machine learning service, a set of candidate variables that can be used to train a model is identified, including at least one processed variable produced by a feature processing transformation. A cost estimate indicative of an effect of implementing the feature processing transformation on a performance metric associated with a prediction goal of the model is determined. Based at least in part on the cost estimate, a feature processing proposal that excludes the feature processing transformation is implemented.

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: An efficient parallel Gibbs sampler using butterfly-patterned partial sums is provided. Instead of building and searching a complete prefix sums table, an alternative “butterfly patterned partial sums table” is described that integrates a lightweight transposition and partial sums operation. Accordingly, the usual full matrix transposition and full prefix sums table building operations can be omitted in favor of building the butterfly-patterned partial sums table, which requires less computational and communication effort. This butterfly-patterned partial sums table is used by a modified binary search phase that calculates the needed prefix-sum table values on-the-fly using the butterfly-patterned partial sums table. Transposed memory access is also provided while avoiding the full matrix transform, providing significant performance benefits for highly parallel architectures, such as graphics processing units (GPUs) where 1-stride or sequential memory accesses are important for optimization.

Abstract: A system and method simulates conversation with a human user. The system and method receive media, convert the media into a system-specific format, and compare the converted media to a vocabulary. The system and method generate a plurality of intents and a plurality of sub-entities and transform them into a pre-defined format. The system and method route intents and the sub-entities to a first selected knowledge engine and a second knowledge engine. The first selected knowledge engine selects the second knowledge engine and each active grammar in the vocabulary uniquely identifies each of the knowledge engines.

Abstract: Described is system for modeling probability matching and loss sensitivity among human subjects. A set of features related to probability matching and loss sensitivity is extracted from collected human responses. The set of features are processed with a genetic algorithm to fit the collected human responses with a set of neural network model instances. A set of model parameters are generated from the genetic algorithm and used to generate at least one of an explanatory and predictive model of human behavior.

Abstract: An embodiment of the invention provides a method for enhanced e-mail return receipts based on cognitive considerations. An input device receives an expected response time from a sender of an electronic message, wherein the expected response time includes the amount of time that the sender expects to receive a response to the electronic message. A processor generates a likelihood that the recipient of the electronic message will respond to the electronic message within the expected response time based on a profile of the recipient. The profile of the recipient includes the recipient's degree of attentiveness to the electronic message, the recipient's workload, the recipient's efficiency, and the recipient's work habits. A communications device presents the likelihood that the recipient will respond to the electronic message within the expected response time to the sender.

Abstract: A system for enhanced geospatial modeling using a spectral data analytic cube classifier and normalized multispectral raster data, comprising a geospatial modeling server that receives and analyzes input imagery, a data import/export server that provides data for review or interaction and receives data to provide to the analysis server, and a database that stores data, and a method for enhanced geospatial modeling using raster data according to the system of the invention.

Abstract: Methods and systems are provided for determining the location of procedure sites, for example hair implantation sites, the method and systems enabling a natural looking randomness to be maintained to achieve a desired density while avoiding previously created procedure sites and existing features. In reference to hair transplantation, methods and systems are provided that allow to account for any terminal and non-terminal hair to assist in selecting locations for hair harvesting or implantation.

Abstract: Content personalized for a user is presented. Particularly, content is personalized and presented to a user in a more cognitive and user-understandable manner to improve the impact and the effectiveness on the user. The system utilizes artificial intelligence to analyze and categorize the content and thereby learns to discover the core concept of the content and any patterns involved. The system also understands the user's interests by capturing the preferred presentation formats and the user's past knowledge. The system maps the categorized content and user's interests and personalizes the content and renders into user preferred presentation type and format. The system supplements the main presentation type with additional related content. The system is capable of continuously monitoring the user activities to understand the effectiveness of the presented content type and formats, and feedback is exploited to continuous improvement of presented content and presentation type and formats.

Abstract: A machine learning service implements programmatic interfaces for a variety of operations on several entity types, such as data sources, statistics, feature processing recipes, models, and aliases. A first request to perform an operation on an instance of a particular entity type is received, and a first job corresponding to the requested operation is inserted in a job queue. Prior to the completion of the first job, a second request to perform another operation is received, where the second operation depends on a result of the operation represented by the first job. A second job, indicating a dependency on the first job, is stored in the job queue. The second job is initiated when the first job completes.

Abstract: A computer-implemented method including: receiving satellite weather data for a macro-location that includes a first micro-location; receiving mobile-sensor data from mobile devices associated with users, the mobile-sensor data sensed by the mobile devices at the first micro-location or a second micro-location included within the macro-location; and forecasting, based on the mobile-sensor data and the satellite weather data, a future micro-climate for the first or the second micro-location.

Abstract: Disclosed are various embodiments for identifying segment assignments of users within segments of interest. Historical user behavior associated with users for whom segment assignments within segments of interest is analyzed. Probabilities associating possible segment assignments within a segment of interest with user behavior are calculated. User behavior of anonymous users and/or users for which segment assignments are unknown can be determined along with a confidence score.

Abstract: Programmatically generating a mixed integer linear programming (“MIP”) matrix, which can then be solved to provide an optimization, based on an annotated entity/relationship data model and a symbolic matrix. The annotated data model identifies one or more outputs of the optimization. The symbolic matrix provides one or more constraints that provide requirements under which the optimization is solved. Outputs of the optimization are represented as variables, inputs of the optimization are represented as constants, and primary keys from the data model are represented as indexes. The constraints are expressed using the variables, constants, and indexes. A MIP matrix is generated from the symbolic matrix, and is then solved by a MIP solver. The output of the MIP solver is used to update a corresponding data structure of the data model.

Abstract: Aspects provide a generic and adaptive approach to adaptive thresholding by using a maximum concentration interval of data to determine one or more adaptive thresholds for any type of operational metric. The generated adaptive thresholds and operational metrics may be used to calculate or otherwise perform a statistical analysis that provides a confidence-level for any changes detected in the operational metric behavior.

Abstract: A self-assembling learning system and apparatus determines self-assembling actions that are expected to reduce uncertainties that are embodied as probabilities and then updates the probabilities in accordance with information that results from the performing of the self-assembling actions. The updated probabilities inform the determination of subsequent self-assembling actions. Neural networks, simulations of multiple potential self-assembling actions, and expected values of information may be applied in determining the self-assembling actions that are to be performed. Sensors may be applied to receive information that inform the updating of probabilities, and the self-assembling apparatus may be a robotic device. Self-assembling actions may comprise modifications to relationships between elements of a computer-implemented system.