Data set III: motor imagery Group information: Department of Biomedical Engineering Tsinghua University Beijing, 100084, P. R. China Supervisor: GAO Xiaorong, Ph.D. ( gxr-dea@tsinghua.edu.cn ) Team Member: JIA Wenyan, Zhao Xianghua Counselor: GAO Shangkai, YANG Fusheng 1. Description of our analysis system [Figure: see DOC file] Processing Flow: a. Selecting subject-specific frequency bands: [10-12Hz] b. Filtering the signals with the bandpass filter c. Computing the energy feature of C3 and C4 for every sampling point e(t)= sum_{i=0,...,n-1} X(t-i)^2 [n: length of window] d. Constructing feature vector (dimension=2) for every trial and for every sampling point E_t= [e_C3(t), e_C4(t)]^T e. Classifying by a linear discriminant analysis (LDA) D_t= W_Tc^T * E_t + w_0 Dt is a time-varying signed distance function (TSD). Dt<0 (Dt>0) means that trial is classified as a left (right) hand trial.