### Table 6: Analogous to the previous table, but for 4-modal orbits. respectively. Moreover, the relationships between symbol sequences of coalescent orbits appear to follow rules analogous to those for symmetric coalescence, with the obvious di erences due to the lack of symmetry. For example, observe from Tables 5 and 6 that a symbol 2 is placed either before or after a string consisting entirely of unchangeable symbols (compare, for example, 3(3,1) and 3(3,5) illustrated in Figures 25 and 26). Also, those containing changeable symbols appear to obey a rule similar to (5.7) (e.g. 4(2,1,6), illustrated in Figure 27).

1994

"... In PAGE 56: ...air (see Section 4.2). The numerical results suggest that the labels of all asymmetric orbits emanating from a bifurcation may be obtained by performing \half quot; the symbol- transformation to the lower-modal symmetric orbit that gives its coalescent partner. For example, observe from Table6 that the orbit 4(2,2,4) bifurcates from (one of) the coalescent pair 3(4,4) and 5(2,2,2,2). For those asymmetric orbits which coalesce, rather than originating in a bifurcation, the behaviour of the orbit with the reversed symbol is implied by the reversibility (as predicted in Section 4.... ..."

Cited by 17

### Table 7: More results on the entire corpus, all use DNASequitur for the grammar inference step. M stands for marker method symbol stream, L stands for LZ77-style symbol stream, KY stands for Kieffer-Yang improvement, and Cost stands for cost measure improvement

2004

### Table 2.3 lists the symbols used for the various domains. Note that there is no symbol for the entire array domain. It is simply denoted by the absence of a domain symbol. The increment of an address order determines how each new address is generated from the previous one. The first increment that springs to mind is +1. But sometimes, other increments are necessary. For pseudo-random address generation for instance, it is necessary to choose a different increment at each run of the march element, to get somewhat random addresses. Table 2.4 lists the possible increments and their corresponding symbols in the address order. When none of the symbols of Table 2.4 is present, +1 increment is the default. Sometimes, it is helpful to address the complement address directly after the current address, before proceeding to the next. The complement address is calculated from an address,

2000

### Table 1. TIMIT symbols

2004

"... In PAGE 5: ... There are 61 symbols used for labeling phonetic segments but most research papers present results on a reduced 39 symbol set. Table1 presents the reduced 39 TIMIT symbol set. 3.... In PAGE 8: ... One way to deal with this problem is to divide the entire phoneme set into phoneme categories by a category classi er and then to recognize phonemes in each category by a phoneme classi er. These categories contain all phonemes from Table1 except silence, because this is the only category having only one phoneme group. We computed the confusion matrix for 64 Gaussian models with 1000 occur- rences per phoneme in training.... ..."

Cited by 1

### Table 6: Our results on the entire corpus: Original Sequitur, DNASequitur with the marker method symbol stream, and DNASequitur with the LZ77-style symbol stream. Size is the number of nucleotides in the original file; all other measurements are in bits per symbol (bps), where 2 bps is the baseline for compression. MPOCPCG is also referred to as CHMPXX; MIPACGA is also referred to as PANMTPACGA

2004

### Table 1: Major Symbols.

1996

"... In PAGE 7: ... 2 Term Weights and Similarity Measure Term weights are determined by the tf idf strategy [15]. ( Table1 summaries the important symbols in this paper.) Precisely, the weight of term j in document i is de ned as: wi;j = tfi;j idfj; where tfi;j is the frequency of occurrence of term j in document i and idfj is the inverse document frequency of term j in the entire collection.... ..."

Cited by 10

### Table 1: Key to symbols and acronyms in the model.

in HP Labs

"... In PAGE 4: ... Examples of such faults include entire-disk or controller failures. We denote the mean time to a visible fault by MV and the associated mean time to repair by MRV (see symbol key in Table1 ). Latent faults are those for which the time between occurrence and detection is sig-... ..."

### Table 1: Key to symbols and acronyms in the model.

"... In PAGE 4: ... Examples of such faults include entire-disk or controller failures. We denote the mean time to a visible fault by MV and the associated mean time to repair by MRV (see symbol key in Table1 ). Latent faults are those for which the time between occurrence and detection is sig- ni cant.... ..."

### Table 1: Key to symbols and acronyms in the model.

in HP Labs

"... In PAGE 4: ... Examples of such faults include entire-disk or controller failures. We denote the mean time to a visible fault by MV and the associated mean time to repair by MRV (see symbol key in Table1 ). Latent faults are those for which the time between occurrence and detection is sig-... ..."

### Table 1: Symbolic LAO* algorithm.

2002

"... In PAGE 4: ... Masking lets us perform dynamic pro- gramming on a subset of the state space instead of the entire state space. The pseudocode in Table1 as- sumes that dynamic programming is performed on E, the states visited by the best (partial) policy. This has been shown to lead to the best performance of LAO*, although a larger or smaller set of states can also be updated (Hansen amp; Zilberstein 2001).... ..."

Cited by 8