Techniques Of Multi-Knowledge Extraction For Brain Cognitive

Functional Magnetic Resonance Imaging (fMRI) is the most important tool for studing the inner workings of the brain cognitive. The data fMRI collected is the specific dynamic patterns generated by different brain regions for specific stimulation things at a time. Obviously, these datas are of large volume and regions which make it a very challenging problem. That is to say, for mining brain behavior changes deeply we must analyze and reduct the collected datas more effectively. So new method which help to abstract useful information from fMRI datas and turn it into comprehensible knowledge effectively and objectively is urgently needed.In this paper, some basic theories about rough set and attribute reduction are studied. Rough set theory is an effective method to reduce the attributes and access rules, but its reduction results are singleness and hardly acquire the minimum reduction. For overcoming these shortcomings, we present a periodic multi-reduction algorithm with formal concept analysis based on rough set. In this algorithm, multi-knowledge is abstracted by multi-reduction of decision table acquired by rough set, incidence of decision values of each condition value in rules acquired by formal concept analysis and then the important rules are obtained. The rules between the single-knowledge base acquired by multi-reduction may exist contradiction which can be determined by bayes classifier and make the multi-knowledge base consistent. Some experiments using UCI datas show the effectiveness of this algorithm. At last, this paper introduce the condition and process of data acquisition fMRI. We obtained the fMRI data through the study of the artificial stimulation of subjects’brain. According to the difference of artificial stimulation type which contains image, Chinese and English this algorithm is applied to different regions in brain to do attribute reduct and then the important rules are obtained by FCA. And last we can judge the artificial stimulation type by the activated degree of each brain region.