Knowledge Amplification Using Randomization in Case-Based Reasoning -- Case Study: Severity of Mammography Mass
Miled Basma Bentaiba-Lagrid, Lydia Bouzar-Benlabiod, Stuart H. Rubin, Thouraya Bouabana‐Tebibel, Maria Roumaissa Hanini · 2018
Case-Based Reasoning (CBR) is the process of resolving new problems based on previous experiences. It relies on the reuse by bringing previously solved problems into its knowledge-base and uses them to solve new ones. Current research trends focus on proposing efficient ways to feed the knowledge bases. Among them, the randomization approach that can feed the knowledge base with less space and fast resolution time. However, the generated knowledge is not necessarily valid. It needs to be validated before their use. In this paper, a new randomization technique to amplify the case-base is presented where the generated data is not necessarily valid. In our method, the validation is done using three layers: coherence verification, stochastic validation and absolute validation. We carried out experiments to predict the severity of a mammography mass. Experiments showed that our approach can significantly improve the resolution efficiency of CBR.