"... In PAGE 6: ...6. Recognition results The recognition results of 63 image sequences are shown in Table4 .... ..."

"... In PAGE 6: ...6. Recognition results The recognition results of 63 image sequences are shown in Table4 .... ..."

"... In PAGE 15: ... Theoretical analysis nonetheless suggests that the CLSM algorithm selects in favor of organisms with rapid learning time and which are thus as small as possible (small size implies fewer learning steps). Cross-sectional data from Simulation 4 (see Table4 ) support the hypothesis that there is no explosion in the size of CLSM organisms as problem complexity increases. The connection/complexity ratio (a/b) in Table 4 provides additional evidence that the size of CLSM organisms with high accuracy of computation does not increase drastically for problems of unusual complexity.... In PAGE 15: ... Cross-sectional data from Simulation 4 (see Table 4) support the hypothesis that there is no explosion in the size of CLSM organisms as problem complexity increases. The connection/complexity ratio (a/b) in Table4 provides additional evidence that the size of CLSM organisms with high accuracy of computation does not increase drastically for problems of unusual complexity. It thus seems legitimate to conjecture that for those problems the CLSM is capable of resolving learning time... ..."

"... In PAGE 15: ...s larger than that of AU12. So we use the angle to represent its orientation if it is present. Although the nose wrinkles are located in the upper face, but we classify the parameter of them in the lower face feature because it is related to the lower face AUs. The definitions of lower face parameters are listed in Table3 . These feature data are affine aligned by calculating them based on the line connected two inner corners of eyes and normalized for individual differences in facial conformation by converting to ratio scores.... In PAGE 24: ....2. Lower Face Action Units Recognition We used a three-layer neural network with one hidden layer to recognize the lower face action units. The inputs of the neural network are the lower face feature parameters shown in Table3 . 7 parameters are used except two parameters of the nasolabial furrows.... ..."

"... In PAGE 15: ...s larger than that of AU12. So we use the angle to represent its orientation if it is present. Although the nose wrinkles are located in the upper face, but we classify the parameter of them in the lower face feature because it is related to the lower face AUs. The definitions of lower face parameters are listed in Table3 . These feature data are affine aligned by calculating them based on the line connected two inner corners of eyes and normalized for individual differences in facial conformation by converting to ratio scores.... In PAGE 24: ....2. Lower Face Action Units Recognition We used a three-layer neural network with one hidden layer to recognize the lower face action units. The inputs of the neural network are the lower face feature parameters shown in Table3 . 7 parameters are used except two parameters of the nasolabial furrows.... ..."

"... In PAGE 6: ... Recall that the main goal of data aggregation is to achieve energy efficiency by transmitting less infor- mation. Table4 shows the number of bytes sent by all three approaches. We observe that no aggrega- tion transmits 75%- and 148% more data than iso- lines in the 16X16 and 32X32 sensor field scenarios, respectively.... ..."

"... In PAGE 6: ... Recall that the main goal of data aggregation is to achieve energy efficiency by transmitting less infor- mation. Table4 shows the number of bytes sent by all three approaches. We observe that no aggrega- tion transmits 75%- and 148% more data than iso- lines in the 16X16 and 32X32 sensor field scenarios, respectively.... ..."