Predicting grid user trustworthiness using neural networks

To addresses the problem of job failures in grid, which might be due to interaction between unknown entities, a reputation based multi agent system is proposed in this paper. The system is based on cooperative model of society in which agents share their experiences about the resource providers through feedback ratings. The uncertainty present in the feedback ratings is handled through Fuzzy Inference System (FIS). The resource providers also compute the trustworthiness of the user before giving access of their resources to safeguard themselves from malicious attacks, using neural networks. The resource providers train the neural network with their own data of already serviced user and predict the trustworthiness of the requesting user. Experiments confirm that the methods with neural networks are feasible and effective for estimation of the trustworthiness of the user.