International Business Machines Corporation

Tibco Software Inc.

Cloud Software Group, Inc.

Daniel Jay Rope

Graham John Wills

T Alan Keahey

Jing-Yun Shyr

Thomas A Hill

Michael A O'Connell

Damir Spisic

Stephen David Gibson

Thomas Alan Keahey

David M Katz

Behrooz Ashtiani

Marc S Altshuller

Margaret J Vais

Michael D Woods

Paolo Angioletti

Gaia Valeria Paolini

Tun-Chieh Hsu

Devendra G Tasgaonkar

Michael Anthony O'Connell

Sivakumar Ramalingam

Jags Ramnarayan

Venkata Jagannath Yellapragada

A set of transition characteristics can be identified. The set of transition characteristics can include continuities and discontinuities between data fields and data visualization channels among a plurality of data visualizations. A transition tuple for each of a plurality of pairs of the data visualizations can be determined and mapped to a matrix. The matrix can be transposed to generate vectors, each vector representing a type of transition characteristic, at least one of the vectors comprising a plurality of elements equaling a total number of the determined transition tuples. The distribution of the data fields and the data visualization channels across the data visualizations can be determined by performing statistical analysis on each of the vectors. A collection of the data visualizations can be determined based on the statistical analysis performed on each of the vectors, the collection of the data visualizations can include a subset of the data visualizations.

Determining a collection of data visualizations

A system for identifying information in high dimensional, low latency streaming data having dynamically evolving data patterns. The system processes, continuously and in real-time, the streaming data. Processing includes filtering the data based on event data to identify diagnostic data points by comparing the event data with an experimental design matrix and performing a modeling operation using the identified diagnostic data points in order to identify efficiently any current and emerging patterns of relationships between at least one outcome variable and predictor variables. The at least one a-priori, pre-designed experimental design matrix is generated based on combinations of the predictor variables and at least one outcome variable. The experimental design matrix is also generated based on at least one of main effects, limitations, constraints, and interaction effects of the predictor variables and combinations.

Adaptive filtering and modeling via adaptive experimental designs to identify emerging data patterns from large volume, high dimensional, high velocity streaming data

Computer-implemented methods and corresponding systems for packaging source code associated with a pipeline into an executable are provided. The methods include parsing a text string that is a textual representation of a pipeline, automatically augmenting one or more operators to the pipeline, generating source code for the pipeline, and packaging the source code into an executable for an external system. The pipeline includes a plurality of operators authored by a user using multiple programming languages to specify a plurality of operations. The one or more operators as automatically augmented include a first operator for persisting output data or metadata associated with a state of a trained model and/or a second operator for generating a monitoring metric for the trained model. The executable may be an executable file, an application, artifact, or a program that is language agnostic and can be executed in an external system using any programming language.

Language agnostic pipeline packager for machine learning

A system for generating algorithmic models comprising a function module to generate a desirability function, an automated machine learning module, and a UI module. The desirability function defines a single desirability value based on an algorithmic model accuracy criteria, criteria for algorithmic model quality, criteria for model fidelity, and criteria for the benefits and cost of model deployment. Specific hard and soft constraints regarding these and other user-defined criteria can also be specified by the user. The automated machine learning module generates an algorithmic model by training the algorithmic model against a model data set, identifying the model with the greatest desirability with respect to all criteria as combined via the desirability function. The UI module generates a user interface to display the overall desirability as well as all model criteria configured by the user. The displayed criteria and desirability are selectable and definable.

Multi-method system for optimal predictive model selection

According to one embodiment, a method for generating a plurality of candidate visualizations. The method may include receiving a scenario description. The method may also include collecting a plurality of expert data using a training system based on the received scenario description. The method may further include generating at least one predictive model based on the collected plurality of expert data in order to execute the at least one generated predictive model during an application of a plurality of genetic algorithms.

Candidate visualization techniques for use with genetic algorithms

An apparatus for estimating analytics and interactive exploration of big data, stored and/or streaming, using approximate query processing is presented. The apparatus comprises a model constructor and a sampler. The model constructor identifies important predictors variables in big data using feature selection, predictor variables, and outcome variables and partitions the important predictor variables into one or more stratifications based either the identified interactions or identified relationships. The sampler generates a subset of data by querying the big data using a query constructed based on at least one stratification. The subset of data can be fed into an analytics generator. The analytics generator generates analytics data for the outcome variables based on the subset of data and an analytics algorithm and a visualization, e.g. an interactive visualization, comprising the outcome variables, the important predictor variables, the stratification, the subset of data, and the analytics data.

System for rapid interactive exploration of big data

A data visualization tool ('tool') configured to generate a visualization for tree data. The tool can calculate a number of hierarchical levels of the tree data. The tool can further subdivide a presentation area for the data visualization into a number of concentric parts. The concentric parts are equivalent to the number of hierarchical levels. Each of the concentric parts corresponds to one of the hierarchical levels. Further, the tool can graphically generate the data visualization for presentation via an electronic display. A first of the concentric parts that corresponds to a highest of the hierarchal levels is an outermost one of the concentric parts in the presentation area. Each successive concentric part inward corresponds to a next lower hierarchical level.

Generating an outside-in hierarchical tree visualization

System, method, and computer program product for measuring transitions between visualizations, the method comprising identifying data fields represented in a first visualization and one or more presentation characteristics for the data fields represented in the first visualization, identifying data fields represented in a second visualization and one or more presentation characteristics for the data fields represented in the second visualization, determining a plurality of transition scores, wherein each transition score represents a difference or similarity between the first and second visualizations, relative to either the identified data fields or the presentation characteristics, and generating a composite measure of transition between the first and second visualizations from the plurality of transition scores.

Measuring transitions between visualizations

A set of transition characteristics can be identified. The set of transition characteristics can include continuities and discontinuities between data fields and data visualization channels among a plurality of data visualizations. The set of transition characteristics can be identified by analyzing the plurality of data visualizations and identifying similarities and differences among the data fields and the data visualization channels. A distribution of the data fields and the data visualization channels across the plurality of data visualizations can be determined. A collection of the data visualizations can be determined based on the distribution of the data fields and the data visualization channels across the plurality of data visualizations. The collection of the data visualizations can include at least a subset of the plurality of data visualizations.

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.

Generation of annotated computerized visualizations with explanations for areas of interest

Techniques are described for genomically defining digital genes encoding data visualization elements and potential incremental changes to the elements as the basis for a genetic selection process for automated generating of data visualizations. In one aspect, a method includes receiving set of input data. The method further includes generating digital genes that genomically define data visualization elements based on the input data, and that define potential incremental changes to the data visualization elements. The method further includes executing a genetic selection process with respect to one or more fitness functions on populations of candidate data visualizations that are based on the genomically defined data visualization elements. The method further includes outputting final data visualization output generated by the genetic selection process.

Growing community of inventors

Reston, VA, United States of America

Daniel Jay Rope