Abstract: At a cloud platform, a class of feed is received for an external feed corresponding to an information source, as are an instruction corresponding to a create operation for the external feed, and a dictionary input corresponding to parameters expected by the information source. The external feed produces a corresponding class of events. At the cloud platform, a handler is selected based on the received class of feed and the received create operation; the input dictionary is transferred to the handler; and the handler generates a unique destination to receive events for the class of events. The handler on the cloud platform generates a unique request to the information source to generate events of the class of feed to the unique destination and sends the request to the information source. Events generated from the information source responsive to the unique request are received at the unique destination.

Abstract: Matrix factorization based gradient compression may be applied to an allreduce operation to improve efficiency including the elimination of unnecessary meta data while maintaining accuracy in training of deep learning (DL) of Artificial Intelligence. This compression may include generating a predetermined matrix and a degree of data compression k as a dimension of the predetermined matrix for a plurality of computing nodes. Each computing node may receive a corresponding matrix of matrices to be allreduced, and each corresponding matrix may be decomposed into a plurality of non-fixed matrices and the predetermined matrix. The plurality of non-fixed matrices may be summed to provide an optimized matrix, which may be multiplied by the predetermined matrix to provide a result matrix. The optimized matrix may be designated as a predetermined matrix. These operations may be repeated until all of the matrices received by the computing nodes have been allreduced.

Abstract: A process trace updating method, system, and computer program product include retrieving, by a computing device, one or more historical executions of a process, receiving, by the computing device, a proposed incremental change, with regard to the process, for a proposed process, updating, by the computing device, the historical execution to build a machine learning model, and generating, by the computing device, a decision and a prediction about execution of the proposed process based upon the machine-learning model.

Abstract: A process trace updating method, system, and computer program product include retrieving, by a computing device, one or more historical executions of a process, receiving, by the computing device, a proposed incremental change, with regard to the process, for a proposed process, updating, by the computing device, the historical execution to build a machine learning model, and generating, by the computing device, a decision and a prediction about execution of the proposed process based upon the machine-learning model.

Abstract: Techniques facilitating causal reasoning for explanation of model predictions are provided. A system can comprise a memory that stores computer executable components and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise an interpretation component that generates one or more explanations of a machine learning model prediction. The one or more explanations can be based on causal relationships determined between feature data of a set of feature data and based on dataset point samples around a trace associated with the causal relationships.

Abstract: Various embodiments migrate a legacy application to a multi-tenant computing environment. In one embodiment, at least one virtualized computing container is instantiated on a host system in a multi-tenant computing environment. An instance of the legacy application is executed within the virtualized computing container. The legacy application having been initially configured to run on premise and serve one tenant at a time. The virtualized computing container securely isolates the executing instance of the legacy application from other executing instances of the legacy application. At least one request received from a first client is sent to the instance of the legacy application executing within the virtualized computing container. The virtualized computing container is quiesced based on at least one quiescing criterion having been satisfied by the instance of the legacy application.

Abstract: A neural network models fragmenting method, system, and computer program product include recursively factoring out common prefixes of models, constructing a hierarchy of decomposed model fragments based on the factoring, and grouping the constructed hierarchy for deployment.

Abstract: Techniques for distributing the training of machine learning models across a plurality of computing devices are presented. An example method includes receiving, from a computing device in a distributed computing environment, a request for a set of outstanding jobs for training part of a machine learning model. A system transmits, to the computing device, information identifying the set of outstanding jobs. The system receives, from the computing device, a selected job for execution on the computing device from the set of outstanding jobs. A chunk of training data associated with the selected job and one or more parameters associated with the selected job may be transmitted to the computing device, and the system may take one or more actions with respect to the chunk of data associated with the selected job based on a response from the computing device.

Abstract: Matrix factorization based gradient compression may be applied to an allreduce operation to improve efficiency including the elimination of unnecessary meta data while maintaining accuracy in training of deep learning (DL) of Artificial Intelligence. This compression may include generating a predetermined matrix and a degree of data compression k as a dimension of the predetermined matrix for a plurality of computing nodes. Each computing node may receive a corresponding matrix of matrices to be allreduced, and each corresponding matrix may be decomposed into a plurality of non-fixed matrices and the predetermined matrix. The plurality of non-fixed matrices may be summed to provide an optimized matrix, which may be multiplied by the predetermined matrix to provide a result matrix. The optimized matrix may be designated as a predetermined matrix. These operations may be repeated until all of the matrices received by the computing nodes have been allreduced.

Abstract: A lifecycle management method, system, and computer program product include coordinating hardware, platform and application-level health checks for framework-independent and application-specific monitoring, failure detection, and recovery, coordinating the hardware, the platform, and the application-level health check by state-specific aggregation of distributed atomic status events, and creating a recovery policy based on the state-specific aggregation of the distributed atomic status events.

Abstract: A lifecycle management method, system, and computer program product include establishing a public key infrastructure (PKI) for end-to-end encryption of control plane and data plane communications by providing encryption between arbitrary components for applicant execution where an interaction pattern is isolated, secure, and a multi-tenant environment.

Abstract: A method, system and computer program product for analyzing multidimensional data are disclosed. In embodiments, the method comprises obtaining an original set of data having a sequential order and multiple original dimensions; selecting a topic-based summarization scheme to summarize the original set of data; and applying the selected topic-based summarization scheme to the original set of data to transform the original set of data into a new set of data having fewer dimensions than the original set of data, while preserving, within a defined measure, the sequential order of the original set of data. In embodiments, the selecting a topic-based summarization scheme includes selecting a plurality of topics, each of the topic representing a set of the original dimensions. In embodiments, the applying the topic-based summarization scheme includes performing dimensionality reduction on the original set of data to transform the original dimensions to the topics.

Abstract: A method of generating process trace clusters includes receiving a set of process traces, where a process trace is an application level event log from a computer system; receiving an identification of node types and edge types of the application event log to generate a heterogeneous information network (HIN) graph, where a temporal node type value is a difference between a current system time and a reference timestamp divided by a user chosen time granularity parameter, performing dimension reduction of the HIN to generate clusters of new event types, generating a set of meta-paths that connect nodes of a same reduced event type in the application event log, calculating similarities between process traces using the new event type clusters, and clustering the set of process traces to generate a plurality of disjoint graph models of processes in the application event log based on the clusters of new event types.

Abstract: A method, system, and recording medium for web Application Programming Interface (API) recommendation, including given a plurality of service description document, gathering a bag of words from the document, generating a service-word matrix based on the bags of words describing the service descriptions, and given a query by a user, recommending a ranked list of services based on the service-word matrix.

Abstract: An action sequence including a plurality of actions and a corresponding input dictionary are obtained. A first container, running an image for a first one of the actions implemented in a first programming language, is instantiated, and the image is executed on the input dictionary to obtain a first action result; the input dictionary is then updated with the first action result to obtain an updated input dictionary. A second container, running an image for a second one of the actions implemented in a second programming language, different than the first programming language, is instantiated, and the image is executed on the updated input dictionary to obtain a second action result; the updated input dictionary is further updated with the second action result to obtain a further updated input dictionary. The input dictionaries are independent of the programming languages.

Abstract: An action sequence including a plurality of actions and a corresponding input dictionary are obtained. A first container, running an image for a first one of the actions implemented in a first programming language, is instantiated, and the image is executed on the input dictionary to obtain a first action result; the input dictionary is then updated with the first action result to obtain an updated input dictionary. A second container, running an image for a second one of the actions implemented in a second programming language, different than the first programming language, is instantiated, and the image is executed on the updated input dictionary to obtain a second action result; the updated input dictionary is further updated with the second action result to obtain a further updated input dictionary. The input dictionaries are independent of the programming languages.

Abstract: Various embodiments migrate a legacy application to a multi-tenant computing environment. In one embodiment, at least one virtualized computing container is instantiated on a host system in a multi-tenant computing environment. An instance of the legacy application is executed within the virtualized computing container. The legacy application having been initially configured to run on premise and serve one tenant at a time. The virtualized computing container securely isolates the executing instance of the legacy application from other executing instances of the legacy application. At least one request received from a first client is sent to the instance of the legacy application executing within the virtualized computing container. The virtualized computing container is quiesced based on at least one quiescing criterion having been satisfied by the instance of the legacy application.

Abstract: Network based service composition with variable distance conditions may be provided. A workflow definition may be received and a network topology may be built based on the workflow definition. Relational data may be received and a network instance may be built based on the network topology and the distance measurement. A plurality of network instances may be built, for example, for different distance conditions. One or more paths may be determined for a pair of services based on one or more of the network instances.

Abstract: Various embodiments migrate a legacy application to a multi-tenant computing environment. In one embodiment, at least one virtualized computing container is instantiated on a host system in a multi-tenant computing environment. An instance of the legacy application is executed within the virtualized computing container. The legacy application having been initially configured to run on premise and serve one tenant at a time. The virtualized computing container securely isolates the executing instance of the legacy application from other executing instances of the legacy application. At least one request received from a first client is sent to the instance of the legacy application executing within the virtualized computing container. The virtualized computing container is quiesced based on at least one quiescing criterion having been satisfied by the instance of the legacy application.

Abstract: An action sequence including a plurality of actions and a corresponding input dictionary are obtained. A first container, running an image for a first one of the actions implemented in a first programming language, is instantiated, and the image is executed on the input dictionary to obtain a first action result; the input dictionary is then updated with the first action result to obtain an updated input dictionary. A second container, running an image for a second one of the actions implemented in a second programming language, different than the first programming language, is instantiated, and the image is executed on the updated input dictionary to obtain a second action result; the updated input dictionary is further updated with the second action result to obtain a further updated input dictionary. The input dictionaries are independent of the programming languages.