Other

Unisys Corporation

Bigml, Inc.

Kodak Alaris Inc.

Eastman-kodak Company

Charles Austin Parker

Susan Mary Jennion

Lois Bridgham Coyne

Adam Ashenfelter

Francisco J Martin

Sarah K Inforzato

Mark Steven Brandt

Michael T Kain

Jose Antonio Ortega

Jos Verwoerd

Frederick W Reed

J Justin Donaldson

Duane J McCrory

Robert Frank Inforzato

John C Waller

Joseph Anthony Manico

Andrew Charles Gallagher

Alexander C Loui

Jiebo Luo

Kevin Michael Gobeyn

Andrew C Blose

Dhiraj Joshi

Anthony Narisi

Wei Jiang

James J Leigh

Christopher John Harrer

Eugene David Pendergraft

Frances Ann Laukagalis

Jane Campbell Mazzagatti

Jeffrey John Wilson

Vinh Ha Le

Poul Petersen

Oscar Rovira

Beatriz Garcia

By Priscilla Wallace, Executive

Antonio Blasco

Dana M Zabilansky

Systems and methods of selecting machine learning models/algorithms for a candidate dataset are disclosed. A computer system may access historical data of a set of algorithms applied to a set of benchmark datasets; select a first algorithm of the set of algorithms; apply the first algorithm to an input dataset to create a model of the input dataset; evaluate and store results of the applying; and add the first algorithm to a set of tried algorithms. The computer system may select a next algorithm of the algorithm set via submodular optimization based on the historical data and the set of tried algorithms; apply the next algorithm to the input dataset; capture a next result based on the applying; add the next result to update the set of tried algorithms; and repeat the submodular optimization. The procedure may continue until a termination condition is reached.

Selection of machine learning algorithms

A non-transitory computer-readable medium including instructions, which when executed by one or more processors of a computing system, causes the computing system to: access a machine learning model m, an input data point P to m, P including one or more features, and a prediction m(P) of m for P; create a set of perturbed input data points Pk from P by selecting a new value for at least one feature of P for each perturbed input data point; obtain a prediction m(Pk) for each of the perturbed input data points; analyze the predictions m(Pk) for the perturbed input data points to determine which features are most influential to the prediction; and output the analysis results to a user.

Prediction characterization for black box machine learning models

A method for automatically selecting thematically representative music is disclosed. A processor is used for using a theme-related keyword to search a keyword-indexed video repository to retrieve videos associated with the theme-related keyword; analyzing the retrieved videos to select videos with music; and extracting music tracks and features from the selected videos. The method further includes selecting representative music related to the theme from the extracted music tracks using the extracted features; and storing the selected representative music in a processor accessible memory.

Automatically selecting thematically representative music

A system and method enables users to selectively expose and optionally monetize their data resources, for example on a web site. Data assets such as datasets and models can be exposed by the proprietor on a public gallery for use by others. Fees may be charged, for example, per new model, or per prediction using a model. Users may selectively expose public datasets or public models while keeping their raw data private.

Predictive modeling and data analysis in a secure shared system

We describe a high-level computational framework especially well suited to parallel operations on large datasets. In a system in accordance with this framework, there is at least one, and generally several, instances of an architecture deployment as further described. We use the term 'architecture deployment' herein to mean a cooperating group of processes together with the hardware on which the processes are executed. This is not to imply a one-to-one association of any process to particular hardware. To the contrary, as detailed below, an architecture deployment may dynamically spawn another deployment as appropriate, including provisioning needed hardware. The active architecture deployments together form a system that dynamically processes jobs requested by a user-customer, in accordance with customer's monetary budget and other criteria, in a robust and automatically scalable environment.

Evolving parallel system to automatically improve the performance of multiple concurrent tasks on large datasets

Systems and processes are disclosed for advanced text analysis in the field of big data analytics and visualization: Users can now factor text into their predictive models, alongside regression, time/date and categorical information. This is ideal for building models where text content may play a prominent role (e.g., social media or customer service logs). Multiple data types, including text fields, may be combined together in datasets and models, and may be presented in various interactive visualization displays.

Interactive visualization of big data sets and models including textual data

Systems and methods are disclosed for building and using decision trees, preferably in a scalable and distributed manner. Our system can be used to create and use classification trees, regression trees, or a combination of regression trees called a gradient boosted regression tree (GBRT). Our system leverages approximate histograms in new ways to process large datasets, or data streams, while limiting inter-process communication bandwidth requirements. Further, in some embodiments, a scalable network of computers or processors is utilized for fast computation of decision trees. Preferably, the network comprises a tree structure of processors, comprising a master node and a plurality of worker nodes or 'workers,' again arranged to limit necessary communications.

Methods for building regression trees in a distributed computing environment

A method for determining a semantic concept associated with an audio signal captured using an audio sensor. A data processor is used to automatically analyze the audio signal using a plurality of semantic concept detectors to determine corresponding preliminary semantic concept detection values, each semantic concept detector being adapted to detect a particular semantic concept. The preliminary semantic concept detection values are analyzed using a joint likelihood model based on predetermined pair-wise likelihoods that particular pairs of semantic concepts co-occur to determine updated semantic concept detection values. One or more semantic concepts are determined based on the updated semantic concept detection values. The semantic concept detectors and the joint likelihood model are trained together with a joint training process using training audio signals, at least some of which are known to be associated with a plurality of semantic concepts.

Audio signal semantic concept classification method

A method for controlling a device responsive to an audio signal captured using an audio sensor. A data processor is used to automatically analyze the audio signal using a plurality of semantic concept detectors to determine corresponding preliminary semantic concept detection values, each semantic concept detector being adapted to detect a particular semantic concept. The preliminary semantic concept detection values are analyzed using a joint likelihood model based on predetermined pair-wise likelihoods that particular pairs of semantic concepts co-occur to determine updated semantic concept detection values. One or more semantic concepts are determined based on the updated semantic concept detection values, and the device is controlled responsive to identified semantic concepts. The semantic concept detectors and the joint likelihood model are trained together with a joint training process using training audio signals, at least some of which are known to be associated with a plurality of semantic concepts.

Audio based control of equipment and systems

A method for detecting, scaling, and positioning handwritten information from hardcopy media, the method includes the steps of digitizing the media and detecting handwritten information; analyzing the media to determine an available location in the digitized image for placement; determining the orientation of the media and handwritten information; and orienting, scaling, and positioning the detected handwritten information. The method further includes storing or using the digitized image side with the handwritten information.

Use of handwritten notations from photographs

A method for producing a slide show video from a collection of hardcopy media, the method includes digitizing the media and detecting handwritten information and estimating the age of the media; determining an order of presentation for the slide show video based on the detected handwritten information and estimated ages; and producing a slide show video from the hardcopy media using the determined order of presentation.

Using handwritten notations in digital video presentations

Methods and apparatus for grouping elements of element pairs into element sets are disclosed. A master processor transmits number pairs corresponding to the element pairs to worker processors, selectively assigns processing of one or more of the number pairs to the worker processors, receives developed sets of numbers from the worker processors, and processes the developed sets of numbers to generate the element sets. Worker processors process the assigned number pairs using a row hash table and a column hash table created from the number pairs corresponding to the element pairs to develop the sets of numbers.

Methods and apparatus for grouping elements of element pairs into element sets

Growing community of inventors

Exton, PA, United States of America

Charles Austin Parker