Font Size: a A A

Language Model, The Access Method And Its Application

Posted on:2008-08-25Degree:MasterType:Thesis
Country:ChinaCandidate:Z H LiuFull Text:PDF
GTID:2208360215485605Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
The fact of making the model obtained by a linguistic Fuzzy Rule-based Model be better highly interpretable involves establishing hard restrictions to the rule structure thus losing flexibility and accuracy, on the contrary, relaxing such restrictions as approximate Fuzzy Rule-Based Model do allows more flexible models to be obtained but their interpretability. So, in fuzzy modeling with Fuzzy Rule-based, we may usually find two contradictory requirements, which are the interpretability and the accuracy of the model obtained. For some complex questions, the people need to not only reflect accurately the rations of system input/output by applying fuzzy model, but also improve the cognition and description characteristic of object by fuzzy modeling and the structure and parameter of model (the process of cognitive modeling). At the same time, as knowledge expression, fuzzy model is needed to have interpretability. This thesis will develop to study interpretable fuzzy model at this background-linguistic fuzzy modeling, trying to acquire the linguistic model with better accuracy not losing the ability of linguistic description.The paper firstly introduces the basic theoretical knowledge of linguitic modeling, and elaborates some basic concepts and basic structure and reasoning mechanism related linguistic model.Secondly, start from the characteristic of linguistic model (behaving well in cognition and description, but the accuracy of linguistic model is not a sufficient degree to a complicated object), developing the research on trade-off between interpretation and accuracy for improving the precision of linguistic model: first, this thesis adopt weighted lingruistic rule using linguistic hedges; second, this thesis introduce triangular-shape membership function, and adopt the method of intelligent learning for selecting linguistic labels; third, simplily the reasoning compute of linguistic model.Finally, combine linguistic model with computational intelligent method, propose a method for learning linguistic model based on the complex intelligent technique of simulated annealing and genetic algorithm, and apply this technique to linguistic modeling for predicting the DAG datum of the transformer and distinguishing from the running state of the transformer. Every model behaves well in accuracy and generalization, and these condition recognition models make up for the defect of IEC way.
Keywords/Search Tags:linguistic model, interpretability, simulated annealing, genetic algorithm, transformer
PDF Full Text Request
Related items