### TABLE II. Non-dominated fuzzy models based on the test data in example A Solutions No. of Rules No. of Sets Rules leng. MSE Train MSE Test

2003

Cited by 1

### Table 2: Cycle time delay breakdown comparison.

1996

Cited by 15

### Table 1: Parameters of the time delay model.

1999

Cited by 1

### Table 2. Weekly variations of the internet time delay.

### Table 4: Some of the results obtained with various time-delay neural networks. Boldface indicates the best result.

2003

"... In PAGE 9: ... Further, at most one layer of hidden units was used, and no coupled weights were used there. Table4 gives some of the results we obtained with time-delay networks. The most obvious feature of this table is how well the combination of categories 4, 5, and 6 again appears to work with this network architecture.... ..."

Cited by 1

### Table 4: Some of the results obtained with various time-delay neural networks. Boldface indicates the best result.

2003

"... In PAGE 16: ... Further, at most one layer of hidden units was used, and no coupled weights were used there. Table4 gives some of the results we obtained with time-delay networks. The most obvious feature of this table is how well the combination of categories 4, 5, and 6 again appears to work with this network architecture.... ..."

Cited by 1

### Table 1. An example data set: samples of one state variable xi, measured every t = 0:001 seconds. dE-dimensional reconstruction-space vectors r(t) from dE time-delayed samples of the xi(t), such that r(t) = [xi(t); xi(t ? ); xi(t ? 2 ); : : : ; xi(t ? (m ? 1) )] or

1999

"... In PAGE 14: ... 3.1 Embedding: the basic ideas Consider a data set comprised of samples xi(t) of a single state variable xi in a n-dimensional system, measured once every t seconds, such as the example sensor time series shown in Table1 . To embed such a data set, one constructs... In PAGE 15: ...Table1 is embedded in two dimensions (dE = 2) with a delay = 0:005, the rst few points in the reconstruction-space trajectory are: (1.6352 1.... ..."

Cited by 2

### Table 2: SIR improvement for the instantaneous, time-delayed and convolutive mixing scenarios (in dBs).

"... In PAGE 3: ...0 dB REL is reduced from 30.78 dB to 26.42 dB. However, this is not really a problem as the SIR of the signal still remains at a high level, so that the speech recognition rate is not affected. Finally, in Table2 and in Figure 2 we present the mean SIR improvement for all three mixing scenarios for both output channels. The overall mean SIR improvement taking all cases into account reaches 11.... ..."