Abstract: Techniques regarding code retrieval tasks 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 comprise a code retrieval component that can execute a code retrieval machine learning task by computing an amount of similarity between neural network embeddings of graph representations of a query text and at least a portion of a computer program code.

Abstract: Embodiments relate to a system, program product, and method for supporting elastic execution of a machine learning (ML) workload using application based profiling. A joint profile comprised of both ML application execution and resource usage data is generated. One or more feature(s) and signature(s) from the joint profile are identified, and a ML execution model for ML application execution and resource usage is built. The ML execution model leverages the feature(s) and signature(s) and is applied to provide one or more directives to subsequent application execution. The application of the ML execution model supports and enables the ML execution to elastically allocate and request one or more resources from a resource management component, with the elastic allocation supporting application execution.

Abstract: Techniques for generating a grounded video description for a video input are provided. Hierarchical Attention based Spatial-Temporal Graph-to-Sequence Learning framework for producing a GVD is provided by generating an initial graph representing a plurality of object features in a plurality of frames of a received video input and generating an implicit graph for the plurality of object features in the plurality of frames using a similarity function. The initial graph and the implicit graph are combined to form a refined graph and the refined graph is processed using attention processes, to generate an attended hierarchical graph of the plurality of object features for the plurality of frames. The grounded video description is generated for the received video input using at least the hierarchical graph of the plurality of features.

Abstract: Techniques regarding code retrieval tasks 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 comprise a code retrieval component that can execute a code retrieval machine learning task by computing an amount of similarity between neural network embeddings of graph representations of a query text and at least a portion of a computer program code.

Abstract: Techniques for generating a grounded video description for a video input are provided. Hierarchical Attention based Spatial-Temporal Graph-to-Sequence Learning framework for producing a GVD is provided by generating an initial graph representing a plurality of object features in a plurality of frames of a received video input and generating an implicit graph for the plurality of object features in the plurality of frames using a similarity function. The initial graph and the implicit graph are combined to form a refined graph and the refined graph is processed using attention processes, to generate an attended hierarchical graph of the plurality of object features for the plurality of frames. The grounded video description is generated for the received video input using at least the hierarchical graph of the plurality of features.

Abstract: Embodiments relate to a system, program product, and method for supporting elastic execution of a machine learning (ML) workload using application based profiling. A joint profile comprised of both ML application execution and resource usage data is generated. One or more feature(s) and signature(s) from the joint profile are identified, and a ML execution model for ML application execution and resource usage is built. The ML execution model leverages the feature(s) and signature(s) and is applied to provide one or more directives to subsequent application execution. The application of the ML execution model supports and enables the ML execution to elastically allocate and request one or more resources from a resource management component, with the elastic allocation supporting application execution.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include partially updating a user's feature when an update ratio is less than a pre-defined threshold ratio, the pre-defined threshold ratio being configurable based on a preference of an existing sparse matrix factorization, where the pre-defined threshold ratio is pre-defined by setting the pre-defined threshold ratio to variable values according to a past acceptable update.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include partially updating a user's feature when an update ratio is less than a pre-defined threshold ratio, the pre-defined threshold ratio being configurable based on a preference of an existing sparse matrix factorization, where the pre-defined threshold ratio is pre-defined by setting the pre-defined threshold ratio to variable values according to a past acceptable update.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include receiving an existing sparse matrix factorization saved in a data store to incrementally update factorized features when users change ratings of the sparse matrix factorization, determining an update ratio based on the number of changed ratings of the sparse matrix factorization, and updating the users' features by using the existing content ratings of the existing sparse matrix factorization if the update ratio is less than a pre-defined threshold ratio.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include receiving an existing sparse matrix factorization saved in a data store to incrementally update factorized features when users change ratings of the sparse matrix factorization, determining an update ratio based on the number of changed ratings of the sparse matrix factorization, and updating the users' features by using the existing content ratings of the existing sparse matrix factorization if the update ratio is less than a pre-defined threshold ratio.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include receiving an existing sparse matrix factorization saved in a data store to incrementally update factorized features when users change ratings of the sparse matrix factorization, determining an update ratio based on the number of changed ratings of the sparse matrix factorization, and updating the users' features by using the existing content ratings of the existing sparse matrix factorization if the update ratio is less than a pre-defined threshold ratio.

Abstract: An alternating least square recommendation method, system, and non-transitory computer readable medium, include receiving an existing sparse matrix factorization saved in a data store to incrementally update factorized features when users change ratings of the sparse matrix factorization, determining an update ratio based on the number of changed ratings of the sparse matrix factorization, and updating the users' features by using the existing content ratings of the existing sparse matrix factorization if the update ratio is less than a pre-defined threshold ratio.

Abstract: The present invention is directed to provisioning and managing computing services in a computing utility system. It receives as an input an infrastructure independent description of a set of requirements on the new desired state of a computing service. It uses a knowledge plane to represent the infrastructure. The method generates a Concrete Model that describes a resource structure that refines the input and is implementable over the infrastructure. It then generates and possibly executes provisioning actions to create an identical resource structure on the infrastructure. The method can be used to create new computing services, to destroy existing computing services, to modify the resource combinations allocated to a computing service, or the configuration of these resources. Provisioning actions can be executed immediately, or saved and executed later, and possibly many times.

Abstract: The present invention is directed to provisioning and managing computing services in a computing utility system. It receives a Concrete Resource Model describing a desired resource structure. It uses a representation of the infrastructure, including resources, configuration, capabilities and constraints, as a knowledge subsystem. The method generates and executes provisioning actions to affect the infrastructure resources by creating a resource structure that matches the description in the Concrete Resource Model and satisfies the requirements described in the input. The method can be used to create new computing services, to destroy existing computing services, to modify the resource combination allocated to a computing service, or the configuration of these resources. Provisioning actions, generated, can be executed immediately, or saved and executed later, and possibly many times.

Abstract: A method, system, and computer program product are disclosed for logically provisioning resources in a data processing system. A request for one or more resources to be provisioned in a provisioning environment is received. The requested resources may be of multiple different available types of resources. Particular instances of the requested types of resources are selected from a group of different types of unassigned available resources. The selected resources are indicated as being reserved, and thus no longer unassigned or available, to fulfill this request. The selected resources are then logically provisioned by establishing logical relationships between the selected resource instances in the provisioning environment.

Abstract: A method, system, and computer program product are described for managing the provisioning of resources in a data processing system. There are different types of resources. Each resource is one of these types. Multiple provisioning states are defined for each type of resource. Relationships are defined among the provisioning states. The relationships describe valid transitions from one state to another. Workflows of one or more tasks are defined, varying to the resource implementation specifications, that are associated with each one of the valid transitions between states. When a resource is to transition from a current state, the relationships among states are used to determine a valid next state. The workflow specific to the resource specification and associated with this transition from the current state to the determined next state is then executed in order to transition to the next state.

Abstract: A generic mechanism and model are provided to capture and store configuration profiles for resource monitors such that monitors can be flexibly deployed, configured, modified, and removed using appropriate sets of parameters and values. As an integral part of resource provisioning, this generic mechanism will require only a generic set of workflows and a generic data model to provision monitors. The provisioning system of the present invention captures and stores response parameters and values from monitors created during initial deployment such that these parameters can be used for later modification and removal of these monitors.

Abstract: A generic method and model for capturing and storing configuration profiles for each resource in a provisioning system. As a resource is utilized in a specific environment, the configuration values can be changed as needed to reflect different stages in the life cycle of the resource, such as configuration, and modification.