B. LEVEL 2: ‘‘MULTIPLE EMBEDDING’’ WITH DIFFERENT (NON-REPEATED) PEAK BINSTo reduce the number of those constraints in empirical schemes (Level 1), our previous work [30] employed multiple embedding method with much more flexible selection approach to determine peak bins, which achieves a better performance. However, some inherent constriction for multiple embedding method could not be completely removed. It requires all the peak and zero bins should be chosen at one time and those values are different from each other, which is defined as ‘‘non-repeated (also called single) peak bins’’ case. The scheme could be illustrated in Fig.4(b) and Fig.5(a).