International Business Machines Corporation

Anne Elizabeth Gattiker

Sani Richard Nassif

Fadi Hikmat Gebara

Rouwaida N Kanj

Anthony N Hylick

Paul Steven Zuchowski

Jody John Van Horn

Thomas S Hubregtsen

Phil Nigh

Franco Motika

Joseph N Kozhaya

Chin Ngai Sze

Samuel Jonathan Thomas

Sujatha Kashyap

Marc Douglas Knox

David A Grosch

Anupama Jagannathan

Minh Ngoc Binh Nguyen

Corville O Allen

Lakshminarayanan Krishnamurthy

Kaipeng Li

Tom Osiecki

Jian Li

Damir Anthony Jamsek

Phillip J Nigh

William A Beason

Karthick Rajamani

Inseok Hwang

Florian Pinel

Jinho Lee

Christopher McCall Durham

William Evan Speight

Steven Frederick Oakland

Donald Lawrence Wheater

Swaminathan Chandrasekaran

Mary Swift

Sridhar Sudarsan

Anita Govindjee

Leah M P Pastel

Thomas Henry Osiecki

Leo A Noel

Mayoore Selvarasa Jaiswal

Raymond J Bulaga

Ambal Balakrishnan

Jody John VanHorn

John L Harris

William J Thibault

Min-Yu Tsai

Matthew Comer

Vincent J Dowling

Provided is a method, computer program product, and system for predicting image sharing decisions using machine learning. A processor may receive a set of annotated images and an associated text input from each user of a plurality of users. The processor may train, using the set of annotated images and the associated text input from each user, a neural network model to output an image sharing decision that is specific to a user.

Image dispositioning using machine learning

A mechanism is provided to implement an action-object interaction detection mechanism for recognizing actions in cluttered video scenes. An object bounding box is computed around an object of interest identified in a corresponding label in an initial frame where the object of interest appears in the frame. The object bounding box is propagated from the initial frame to a subsequent frame. For the initial frame and the subsequent frame: the object bounding boxes of the initial frame and the subsequent frame are refined and cropped based on the associated refined object bounding boxes. The set of cropped frames are processed to determine a probability that an action that is to be verified from the corresponding label is being performed. Responsive to determining the probability is equal to or exceeds a verification threshold, a confirmation is provided that the action-object interaction video performs the action that is to be verified.

Action-object recognition in cluttered video scenes using text

Examples of techniques for constructing, evaluating, and improving a search string for retrieving images are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes receiving, by a processing device, a plurality of images as search results returned based at least in part on a search string for an item in the form of a tuple including an item class, an action and an actor. The method further includes determining, by the processing device, whether the search string is effective at indicating a common item use based on image similarity. The method further includes, based at least in part on determining that the search string is ineffective at indicating the item use, generating, by the processing device, an alternative search string.

Constructing, evaluating, and improving a search string for retrieving images indicating item use

An object dataset creation or modification mechanism is provided for object dataset creation or modification using a labeled action-object video. For a plurality of frames of the labeled action-object video, an identification is made of a subset of frames where a bounding box object (BBO) exists. BBOs in the subset of frames where a BBO exists are pruned to identify sufficiently distinct BBOs thereby forming a set of pruned BBOs. For each pruned BBO in the set of pruned BBOs: an information addition score is determined; the information addition score is assessed; responsive to the information addition score being positively assessed, the pruned BBO is added to an object dataset; and, responsive to the information addition score being negatively assessed, the pruned BBO is discarded.

Object dataset creation or modification using labeled action-object videos

A method for determining an intervention response category for provisioning workflows. The method determines provisioning features of a provisioning step. The method performs outlier detection to identify and remove outliers from non-intervention data to produce a non-intervention normal data set. The method performs iterative grouping on the non-intervention normal data set to determine significant variables in the provisioning features. The method performs response mapping of provisions using results of the iterative grouping including a significance of errors and a presence of errors in the non-intervention normal data and partial intervention data to categorize the provisions into a response category.

Method for proactive trouble-shooting of provisioning workflows for efficient cloud operations

A method, system and computer-usable medium are disclosed for using table-based groundtruth to train a question-answer (QA) system, comprising: receiving a corpus of text comprising unstructured text and a table; performing QA pair generation operations on the unstructured text and the table to generate QA pairs, the QA pairs being based upon information obtained from the table; assigning a score to each QA pair, the score corresponding to an indicator of user interest.

Automating table-based groundtruth generation

Examples of techniques for constructing, evaluating, and improving a search string for retrieving images are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes receiving, by a processing device, a plurality of images returned as results to an executed search string in the form of a tuple including an item class, an action and an actor. The method further includes generating, by the processing device, a collection similarity score by comparing at least one of the plurality of images to at least one other of the plurality of images. The method further includes determining, by the processing device, whether the collection similarity score exceeds a collection threshold. The method further includes, based at least in part on determining that the similarity score does not exceed the collection threshold, indicating that the search string is ineffective at indicating the item use.

Examples of techniques for constructing, evaluating, and improving a search string for retrieving images are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes receiving, by a processing device, an item identifier. The method further includes retrieving, by the processing device, an item description based at least in part on the item identifier. The method further includes identifying, by the processing device, a tuple indicating a common item use based at least in part on the item description. The method further includes constructing, by the processing device, a search string based at least in part on the tuple. The method further includes retrieving, by the processing device, at least one image based at least in part on the search string.

A system and a computer program product are provided for evaluating question-answer pairs in an answer key by comparing a first answer key answer to a plurality of candidate answers to determine if the answer key may have a problem if the plurality of candidate answers are more similar to one another than to the first answer and to determine if the plurality of candidate answers has gradient information which may be used to update the answer key if not already included in the answer key.

System and method for identifying answer key problems in a natural language question and answering system

A method and system are provided for converting clinical criteria constraints for a target structure of radiation therapy into a second set of constraints. The method includes selecting, by a hardware processor, a plurality of clusters of voxels in the target structure based on pre-specified criteria. The method further includes assigning, by the hardware processor, each of the plurality of clusters of voxels a respective constraint that specifies one or more bounds on a radiation dose applied to each voxel in a corresponding one of the plurality of clusters of voxels. The method also includes storing, in a memory device, the respective constraint for each of the plurality of clusters of voxels.

Translating different clinical protocols for particle therapy into a set of constraints

Examples of techniques for constructing, evaluating, and improving a search string for retrieving images are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes constructing, by a processing device, a search string based at least in part on a tuple including an item class, an action, and an actor. The method further includes retrieving, by the processing device, a plurality of images based at least in part on the search string for an item. The method further includes evaluating, by the processing device, the retrieved plurality of images based on a similarity to determine whether the search string is effective at indicating a common item use. The method further includes, based at least in part on determining that the search string is ineffective at indicating the item use, generating, by the processing device, an alternative search string.

An apparatus comprising a memory and a processor coupled to the memory. The processor receives input from a user, processes a first portion of the input via more than one service module while receiving a second portion of the input to determine a first speculative result, wherein processing the first portion of the input comprises executing at least one service module coupled to a corresponding speculation buffer, processes a second portion of the input via the more than one service module to determine a second speculative result, wherein processing the second portion of the input comprises executing the at least one service module coupled to the corresponding speculation buffer, processes the input via the more than one service module to determine a final output, wherein processing the input comprises executing the at least one service module coupled to the corresponding speculation buffer, and outputs the final output to the user.

Pre-processing partial inputs for accelerating automatic dialog response

A method, system and computer-usable medium are disclosed for automating the generation of table-based groundtruth, comprising: receiving a document comprising unstructured text and a table; generating questions by applying a template the contents of the table; performing QA pair generation operations on the table to generate QA pairs, each QA pair comprising a question generated by applying the template; and, assigning a score to each QA pair, the score providing an indicator of user interest to each QA pair, the score being based on a score generation methodology using the unstructured text and the table.

A computer-implemented method for creating question-answer pairs is provided. The computer-implemented method includes leveraging domain specific resources including, at least one or more of lexicons, glossaries, or knowledge bases for constructing templates for creating the question-answer pairs. The computer implemented method further includes leveraging user experiences of a plurality of users for constructing templates. The computer implemented method further includes eliminating erroneous question-answer pairs based on templates specifications of a heuristic process of the constructed templates.

Domain-specific question-answer pair generation

Growing community of inventors

Austin, TX, United States of America

Anne Elizabeth Gattiker