International Business Machines Corporation

Seraphin Bernard Calo

Dinesh C Verma

Douglas Morgan Freimuth

Erich Miles Nahum

Kang-Won Lee

Maroun Touma

Bong Jun Ko

Raheleh B Dilmaghani

Vasileios Pappas

Thai Franck Le

Christian Makaya

Ramya Raghavendra

Theodoros Salonidis

Franck Vinh Le

Dakshi Agrawal

Raghu Kiran Ganti

Fan Ye

William Cornejo

Yang Song

Xiping Wang

David Alson George

Keith William Grueneberg

Thai V Le

Mandis S Beigi

David Louis Kaminsky

Renu Tewari

Anees Abdul Shaikh

Casimer M DeCusatis

Ramani Ranjan Routray

Kaladhar Voruganti

Murthy V Devarakonda

Sandeep Gopisetty

Norbert George Vogl

James J Fan

Supriyo Chakraborty

Murtaza Zafer

Chung-hao Tan

Shiqiang Wang

David Alvra Wood, Iii

Jorge J Ortiz

Konstantinos Magoutis

Ho Yin Starsky Wong

Gauri Sudhir Shah

Wei-Han Lee

Abm Musa

Gary Lipson

Mudhakar Srivatsa

Mukesh Kumar Mohania

Chatschik Bisdikian

Steve Richard White

Petros Zerfos

James R Giles

Nirmit V Desai

David R Vyvyan

Tova Roth

Dilip Dinkar Kandlur

Jae-Wook Ahn

Jorge Lobo

David A Wood

Richard Orville LaMaire

Geeth Ranmal De Mel

Steven A Waite

Krishnamurthi Kannan

Carolyn A Brodie

Changchang Liu

Thomas G Burket

Kisung Lee

Thomas J C Berman

Ashish Mehra

Suk Suet Soo

Clare-Marie Karat

Eric Nahum

Allan J Finkel

Anastasios T Bouloutas

John W Wiley, Jr

Jiawei Wen

Roger Hollander

Anshu Kak

Lu Su

Joseph D Ryan

Kyoungsoo Park

Techniques for determining system performance without ground truth include receiving a trained model and one or more generator models, the trained model having been trained on training data. The trained model is used on testing data to produce labeled testing data, and the labeled testing data are used to train a proxy model. The one or more generator models are used to produce synthetic training data that are representative of the training data. The proxy model is used on the synthetic training data to produce predictions, and performance of the trained model is determined based on the predictions by the proxy model.

Determining system performance without ground truth

Methods and systems for emulating an application include generating a log template to match one or more patterns in a set of application logs collected from an original application. Semantic state representations are learned for the original application from the log template. A classifier is trained to predict a next action template based on a sequence of prior action templates. A regressor is trained to generate a parameter value for a template based on a sequence of prior action templates and a particular semantic state of the original application.

Machine learning to emulate software applications

A method and system of training an interpretable deep learning model includes receiving an input set of data, which may be complex. The input set of data is provided to deep learning model for feature extraction. In an exemplary embodiment, the deep learning model generates a disentangled latent space of features from the feature extraction. The features may comprise semantically meaningful data which is then provided to a low-complexity learning model. The low-complexity learning model generates output based on a specified task (for example, classification or regression). Being a low-complexity learning model provides confidence that the data output from the deep learning model is inherently interpretable.

Using disentangled learning to train an interpretable deep learning model

A network function optimization method, system, and computer program product include optimizing network function chain components of a software by modifying a structure of the network function chain components by removing a function of the network function chain components.

System, method and computer program product for network function optimization based on locality and function type

A computer-implemented method, a computer program product, and a computer system for enhanced distributed machine learning. A fusion server in a distributed machine learning system determines correlation relationships across agents in the distributed machine learning system, based on auxiliary information. The fusion server clusters the agents to form one or more communities, based on the correlation relationships. The fusion server selects, from the one or more communities, participating agents that participate in the enhanced distributed machine learning.

Leveraging correlation across agents for enhanced distributed machine learning

A computing system for learning a device type and message formats used by a device is provided. The computing system includes an interface and a processor. The interface is receptive of documents describing identification information and communication and application protocols of devices. The processor is coupled with the interface to obtain rules of network packet analysis using document analytics and identify identification information and communication and application protocols of network messages from devices using the rules.

Automatic protocol discovery using text analytics

Regulating a flow of data from an electronic device comprises generating a user profile associated with a user. Further, a resource profile is generated and is associated with one or more resources external to the electronic device. Additionally, a context profile is generated and describes an outcome of one or more previous interactions between the user of the electronic device and one or more resources. Rules are generated based on a comparison the user profile, the context profile and the site profile. Further, data sent from an electronic device is modified based on the set of rules.

Personal advisor for managing private information

A network function optimization method, system, and computer program product, include optimizing network function chain components by modifying a structure of the network function chain components by removing one of the functions of the network function chain components in response to a constraint according to a policy requirement.

Embodiments of the present disclosure include a method, computer program product, and system for determining to push a data packet to a device. A processor may receive a first data packet. The processor may execute the first data packet in a secure environment. The secure environment may simulate a first state of a device. The device may include a firewall. The processor may determine, from the execution of the first data packet, that the first state changed to a second state. The processor may identify that the second state is a predetermined secure state. The processor may push the data packet to the device in response to identifying that the second state is the predetermined secure state.

Cognitive stateful firewall for IoT devices

A method is provided of using a set of servers to provide deferential services that have a pre-negotiated time for notice to release the servers. The method includes defining a virtual checkpoint frame interval that is constrained to a duration of up to half of the pre-negotiated time for notice to release the servers. The method includes, responsive to an end of the interval, (i) writing, to a shared state database, a state of processing of the packets and transactions occurring during the interval that are processed by a current one of the servers, and (ii) releasing the packets and transactions occurring during the interval. The method includes copying the packets and transactions occurring during the interval, and the state, from the current server to another server for subsequent processing, responsive to an indication of an instance loss on the current server.

Synchronized release of resources used in deferential cloud services

A computer program product is provided for fault diagnostics and recovery of network service chains. The product includes a computer readable storage medium having program instructions embodied therewith. The program instructions are executable by a computer to cause the computer to perform a method. The method includes identifying intersecting ones of components of the network service chains in a network graph indicative of a commonality therebetween. The method further includes selectively repairing one or more of the components of at least one of the network service chains up to an entirety thereof based on a component repair sequence that maintains service continuity of the network service chains and that is determined and selectively reordered based on the commonality. The network graph includes a plurality of graphs, each (i) including one or more Virtualized Network Functions, and (ii) corresponding to a respective one of the network service chains.

Mechanism for fault diagnosis and recovery of network service chains

A computer-implemented method is provided for fault diagnostics and recovery of network service chains. The method includes identifying, by a processor, intersecting ones of components of the network service chains in a network graph indicative of a commonality therebetween. The network graph represents the components of the network service chains. The method further includes selectively repairing, by the processor, one or more of the components of at least one of the network service chains up to an entirety thereof based on a component repair sequence that maintains service continuity of the network service chains and that is determined and selectively reordered based on the commonality. The network graph includes a plurality of graphs, each (i) including one or more Virtualized Network Functions, and (ii) corresponding to a respective one of the network service chains.

A method includes configuring worker services to operate in a stateless manner and providing support services that enable the worker services to operate in the stateless manner. The support services include (i) a management service for providing notifications of server removal and addition, (ii) a state maintenance service for maintaining state information in a central location, and (iii) a load balancer service for distributing requests among worker services. The method includes altering a number of servers allocated to at least one worker service, responsive to a notification from the management service. A private protocol is used between the worker services and load balancer service (a) to send, from the worker services to the load balancer service, a respective pointer to the state information associated with the requests, and (b) to include the respective pointer in the requests when any of the requests are forwarded to any worker service.

Deferential support of request driven cloud services

A method is provided of using a set of servers to provide deferential services that have a pre-negotiated time for notice to release the servers. The method includes defining a virtual checkpoint frame interval that is constrained to a duration of up to half of the pre-negotiated time for notice to release the servers. The method includes collecting packets and transactions occurring during the interval that are processed by a current server. The method includes, responsive to an end of the interval, (i) writing, to a shared state database, a state of processing of the packets and transactions occurring during the interval, and (ii) releasing the packets and transactions occurring during the interval. The method includes copying the packets and transactions occurring during the interval, and the state, from the current server to another server for subsequent processing, responsive to an indication of an instance loss on the current server.

Growing community of inventors

Cortlandt Manor, NY, United States of America

Seraphin Bernard Calo