Abstract: Example implementations relate to method and system for implementation of a software agent in a computing system, to identify the most effective self-healing action to build a Q-table by applying a reinforcement learning technique. In particular, the method includes determining an induced state of the computing system based on an error in one or more components of the computing system and selecting a corrective action corresponding to the induced state based on a plurality of Q-values stored in the Q-table. The method further includes executing the corrective action in the computing system and evaluating one or more parameters of the computing system to determine a current state of the computing system. Further, the method includes updating a Q-value of the plurality of Q-values corresponding to the corrective action, in response to evaluation of the current state of the computing system, to build the Q-table.

Abstract: Example implementations relate to method and system for implementation of a software agent in a computing system, to identify the most effective self-healing action to build a Q-table by applying a reinforcement learning technique. In particular, the method includes determining an induced state of the computing system based on an error in one or more components of the computing system and selecting a corrective action corresponding to the induced state based on a plurality of Q-values stored in the Q-table. The method further includes executing the corrective action in the computing system and evaluating one or more parameters of the computing system to determine a current state of the computing system. Further, the method includes updating a Q-value of the plurality of Q-values corresponding to the corrective action, in response to evaluation of the current state of the computing system, to build the Q-table.

Abstract: A computing device includes a processor and a medium storing instructions executable to: detect a performance degradation of a first component of a computing system; in response to a detection of the performance degradation of the first component, filter a plurality of system events of the computing system using an impact matrix to generate a filtered set of system events, wherein each system event of the filtered set is associated with a first set of components of the computing system, wherein the impact matrix indicates one or more components of the first set that can have a performance impact on the first component; perform a linear regression on the filtered set of system events; and generate a ranked list of system events based on the linear regression, the system events in the ranked list being ordered according to likelihood of having caused the performance degradation.