Abstract-To recognize an object in an image, the user must apply a combination of operators, where each operator has a set of parameters. These parameters must be "well" adjusted in order to reach good results. Usually, this adjustment is made manually by the user. In this paper we propose a new method to automate the process of parameter adjustment for an object recognition task. Our method is based on reinforcement learning, we use two types of agents: User Agent that gives the necessary information and Parameter Agent that adjusts the parameters of each operator. Due to the nature of reinforcement learning the results do not depend only on the system characteristics but also the user's favorite choices.

Selecting the appropriate operators with the optimal values for their parameters represents a big challenge for users. In this paper we present a solution for this problem. This solution uses a multi-agent architecture based on reinforcement learning to automate the process of operator selection and parameter adjustment. The architecture consists of three types of agents: the User Agent, the Operator Agent and the Parameter Agent. The User Agent determines the phases of treatment, and for each phase it determines a library of possible operators and possible values of their parameters. The Operator Agent constructs all possible combinations of operators and decides for the best one. The Parameter Agent, the core of the architecture, adjusts the parameters of each combination of operators by processing a large number of images. Towards the end, the agents must offer the best combination of operators and the best values of their parameters.

No abstract

No abstract

No abstract