International Business Machines Corporation

Michael James Hind

Peter Francis Sweeney

Matthew R Arnold

Kush Raj Varshney

Roman Vaculin

Sekou Lionel Remy

Nelson Kibichii Bore

Eleftheria K Pissadaki

Ravi Kiran Raman

Stephen Fink

David Paul Grove

Manish Satyapal Gupta

Sameep Mehta

Ramya Raghavendra

Kalapriya Kannan

Evelyn Duesterwald

Jiri Navratil

John Thomas Richards

Jong-Deok Choi

Aleksandra Mojsilovic

Darrell Christopher Reimer

Rachel K E Bellamy

Manish Kesarwani

Kaoutar El Maghraoui

Michael George Burke

Murray Scott Campbell

Stephanie Houde

David John Piorkowski

Paul Robert Carini

Jason Tsay

Benjamin Tyler Elder

Anupama Murthi

Jeremy Lau

John Whaley

A computer implemented method of performing large-scale machine learning experiments includes expanding on one or more input datasets by systematically generating several data set drift splits. A set of experimental jobs corresponding to the generated data set drift splits are executed to generate experimental results. The experimental results are processed, consolidated, and clustered according to the generated data set drift splits.

Assistant for large-scale machine learning experiments

An example operation may include one or more of generating a plurality of successive data points of an iterative simulation based on predefined checkpoints, each data point identifying an evolving state of the iterative simulation with respect to a previous data point among the successive data points, transmitting a blockchain request for validating state data within a first data point among the plurality of successive data points to a first subset of endorsing nodes of a blockchain network, and transmitting a blockchain request for validating state data within a second data point among the plurality of successive data points to a second subset of endorsing nodes which are mutually exclusive from the first subset of endorsing nodes of the blockchain network for parallel endorsement of the first and second data points.

Distributed platform for computation and trusted validation

An example operation may include one or more of obtaining data of a simulation, identifying checkpoints within the simulation data, generating a plurality of sequential data structures based on the identified checkpoints, where each data structure identifies an evolving state of the simulation with respect to a previous data structure among the sequential data structures, and transmitting the generated sequential data structures to nodes of a blockchain network for inclusion in one or more data blocks within a hash-linked chain of data blocks.

Techniques regarding explanations for artificial intelligence recommendations are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can include: a combination component that receives a first training dataset comprising first feature vectors, first classes and first explanations, and combines the first classes and the first explanations to produce first augmented labels and a second training dataset that comprises the first feature vectors and the first augmented labels; a classifier, trained on the second training dataset, that analyses second feature vectors and generates second augmented labels; and a decomposing component that decomposes the second augmented labels, using the classifier, to generate second classes and second explanations.

Explanations for artificial intelligence based recommendations

A method for automatically generating documentation for an artificial intelligence model includes receiving, by a computing device, an artificial intelligence model. The computing device accesses a model facts policy that indicates data to be collected for artificial intelligence models. The computing device collects artificial intelligence model facts regarding the artificial intelligence model according to the model facts policy. The computing device accesses a factsheet template. The factsheet template provides a schema for an artificial intelligence model factsheet for the artificial intelligence model. The computing device populates the artificial intelligence model factsheet using the factsheet template with the artificial intelligence model facts related to the artificial intelligence model.

Generation and management of an artificial intelligence (AI) model documentation throughout its life cycle

An example operation may include one or more of generating a data frame storing content of a simulation, compressing the simulation content within the data frame based on previous simulation content stored in another data frame to generate a compressed data frame, and transmitting the compressed data frame via a blockchain request to one or more endorsing peer nodes of a blockchain network for inclusion of the compressed data frame within a hash-linked chain of blocks of the blockchain network.

There is provided an autonomic software system and method for normalizing a profile collected for an executing application to account for one or more actions applied to the executing application after the profile was collected, comprising: predicting an impact of applying the one or more actions to the executing application by utilizing the profile and the one or more actions; and adjusting the profile to form a normalized profile according to the predicted impact. A plurality of different a profile consumers, such as, a phase shift detector, an action evaluator as well as a normalizing controller, may utilize the normalized profile to improve the behavior of the executing application. In addition, online visualization tools may be implemented to graphically depict the normalized profiles, as well as differences between the collected profiles and the normalized profiles.

Profile normalization in an autonomic software system

A method for evaluating software performance includes steps of: receiving a plurality of versions of code; selecting starting and stopping points for timing execution of the code versions; dispatching at least two of the plurality of code versions for execution; repeatedly executing the at least two code versions; recording execution times for the at least two code versions, according to the selected starting and stopping points; collecting the execution times; and processing the collected execution times. The method further includes steps of: performing a statistical analysis of the collected execution times for determining which code version is fastest; and invoking a confidence metric periodically to determine if a difference between means is statistically meaningful.

Technique for evaluating software performance online to support online tuning

There is provided an autonomic software system and method for normalizing a profile collected for an executing application to account for one or more actions applied to the executing application after the profile was collected, comprising: predicting an impact on the profile of applying the one or more actions to the executing application; and adjusting the profile to form a normalized profile according to the predicted impact. A plurality of different a profile consumers, such as, a phase shift detector, an action evaluator as well as a normalizing controller, may utilize the normalized profile to improve the behavior of the executing application. In addition, online visualization tools may be implemented to graphically depict the normalized profiles, as well as differences between the collected profiles and the normalized profiles.

Method and system for profile normalization in an autonomic software system

A system and method for characterizing runtime behavior of a computer program executing in an execution environment, the method comprising: identifying one or more instances of yield points in a program to be executed, each yield point indicating a potential sampling operation during program execution; during program execution, in response to an identified yield point instance, ascertaining a state of the execution environment for indicating whether a sampling operation is to be performed; and, when the state of the execution environment indicates a sampling operation, recording relevant information for characterizing behavior of the execution environment. Relevant information for characterizing program behavior includes frequencies of methods executed in the program, and calling context associated with methods called by the program. Different mechanisms are provided for determining the sampling condition including the setting of a trigger bit by a runtime system, or, determining a sampling operations based on a fixed percentage of all executed yield points taken.

System and method for characterizing program behavior by sampling at selected program points

A sampling-based system and method for adaptively optimizing a computer program executing in an execution environment that comprises one or more compiler devices for providing various levels of program optimization. The system comprises a runtime measurements sub-system for monitoring execution of the computer program to be optimized, the monitoring including obtaining raw profile data samples and characterizing the raw profile data; a controller device for receiving the characterized raw profile data from the runtime measurements sub-system and analyzing the data for determining whether a level of program optimization for the executing program is to be performed by a compiler device, the controller generating a compilation plan in accordance with a determined level of optimization; and, a recompilation sub-system for receiving a compilation plan from the controller and invoking a compiler device for performing the level of program optimization of the executing program in accordance with the compilation plan.

System and method for adaptively optimizing program execution by sampling at selected program points

A method for characterizing runtime behavior of a computer program executing in an execution environment comprising: generating a call stack runtime data structure for tracking methods currently active in an executing program thread, an active method on the call stack is represented by a frame; determining condition for sampling an executing program to determine current program behavior; and, upon determination of a sampling condition, the sampling including examining at least one frame in the call stack in response to evaluate context of one or more methods represented in the call stack, the at least one frame in the call stack providing context relating to an executing program's calling structure.

Method for characterizing program execution by periodic call stack inspection

A method and several variants are provided for analyzing and transforming a computer program such that instructions may be reordered even across instructions that may throw an exception, while strictly preserving the precise exception semantics of the original program. The method uses program analysis to identify the subset of program state that needs to be preserved if an exception is thrown. Furthermore, the method performs a program transformation that allows dependence constraints among potentially excepting instructions to be completely ignored while applying program optimizations. This transformation does not require any special hardware support, and requires a compensation code to be executed only if an exception is thrown, i.e., no additional instructions need to be executed if an exception is not thrown. Variants of the method show how one or several of the features of the method may be performed.

Growing community of inventors

Ulster County, NY, United States of America

Michael James Hind